Jump to content

Talk:Wang Laboratories/Archives/2013

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia


Clarifications

I'm not sure how to fit this information into the existing discussions so I thought I would create a new thread. I apologize for the length but wanted to accurately present the information.

Mini-computer - The 2200 was not Wang's first mini computer. Wang had developed a computer called the 4000 around 1967 or 1968. The second foray into that arena, late in 1968, was the 3300 BASIC, which was to compete with the DEC PDP-8 and DG-10. Like its competition, the 3300 was a multi-user CPU with connected teletypewriter terminals. The 3300 was optimized for running BASIC. As was common in similar systems, you had to enter a boot-strap-loader program by means of front panel switches, which then allowed for the reading of programs from paper tape on the teletypewriter. This process could take up to 45 minutes to get the machine operational. By 1973 the 3300 was modified to read data from cassette tapes but by then it had lost its market window and was replaced with the 2200.

The Wang VS was originally thought of as an extension of the 2200 product line and up to the official announcement was referred to as the 8300 within R&D and 2200VS in Marketing. When announced the name was changed to the WCS 60 and WCS 80, the WCS standing for Wang Computer System. When announced and initially shipped, the VS did not support 928 style workstations but instead workstations (2246P) with custom CPUs connected to the VS with parallel cables. Even when 928 workstations became available on the VS, the requirement for at least one parallel workstation remained for some time.

Word Processing/OIS - Wang Word Processing began with the 1220 in 1970. This was an IBM Selectric based typewriter, interfaced with a CPU built out of the 600 series Wang Calculator with dual cassette tapes for storage of documents. This was followed by the 1222 in 1973 which added an Intel 4 bit microprocessor and switched the typewriter to a IBM Selectric II. The 1222 was later modified with the addition of a CRT that allowed the operator to view two lines of text in memory before printing.

In 1975 as Koplow and Morros were writing their Word Processing Operators manual there was already a hardware design team at work on what was intended to be an incremental upgrade to the 1200 line called the 1250. This upgrade would consist of replacing the cassette tapes with 8” floppy diskettes and a new output device to replace the typewriter. As the Koplow Operators Manual took shape, the planned hardware upgrade design was scraped in favor of using another hardware project that was developing an Intel 8080 microprocessor based terminal. The new WP was to connect to the central storage and control unit by parallel cables. In part because of static problems encountered on the 1222, on September 28, 1975, the workstation connection method became serial dual coaxial cables driven in a differential mode, which became known as 928.

The WPS 10, 20 & 30 were announced in June 1976, and shipped with Intel 8080 CPUs. The workstation contained 16K of memory and the printer 8K. As the software evolved, the printer memory was upgraded to 12K.

The 928 Intel 8080 hardware platform was intended as a general purpose file server with diskless workstations attached. In parallel with the WP Software development on this platform was a small group developing a Key to Disk data entry system that was dropped after the WP Product initial shipments began so the concentration could be focused on WP.

The WP systems were initially shipped with narrow focus micro-code in one E-PROM on the Master unit that allowed for reading and writing the disks and communication with the attached devices. This micro-code was later replaced with two E-PROMs that contained an event driven interface supporting a general purpose Disk Operating System.

In early 1978 Wang announced the OIS-130, which was built on the same hardware as the WPS 30 but with workstation memory upgraded to 32K and printers upgraded to 16K of memory. The OIS under its new DOS operating system had the same WP software but added the Basic Language and some new utilities. In October 1978, Wang announced the OIS-140, which was based on the Zilog Z-80 CPU and supported SMD/CMD disk drives as well as more workstations.

With the introduction of Basic on the OIS, a set of Document Access Subroutines was released allowing for customers or third party developers to extend the functionality of the system. Using these same Subroutines, Wang developed and released several add-on packages including List Processing.

OIS workstations kept evolving until the original 3 circuit board 8080 32K unit was replaced with a two board 64K Z-80.

OIS Limitations - It was clear that the OIS was a collection of applications from different development groups lacking not only a common look and feel but also true integration between them. Additionally, users and third party developers wanted the WP Glossary functionality across the entire system not just within WP. Customers were also looking for more sophisticated security.

Alliance 250 was announced in the fall of 1981 with initial shipments in the spring of 1982. The Alliance system utilized the same hardware as the OIS-140. The Alliance software system was built on the same DOS operating system as the OIS but with a new set of APIs called SSR or Slave Service Routines between the application and OS. This gave the applications a common look and feel as well as allowing for a system wide macro capability (Old WP Glossary with many enhancements). Additionally, much of the system was built on an internally developed relational database. The system had a central database that kept track of: the installed Software, Users, Classes, Access Control Lists, etc. Many of the new integrated applications, such as Calendar, e-Mail, etc., also used the same type of database for their functionality. Alliance also introduced an end user relational database, built on this same technology, called Visual Memory. All applications supported some form of dynamic data exchange e.g., a Visual Memory field could be used in the WP document for addressing or a document could be e-mailed to another user. --Dr Glossary 21:44, 27 February 2006 (UTC)

Missing Miscellany

We seem to be missing a few things. "Wangnet", the cctv based broadband LAN, The Laptop where Wang decided that there was no market for laptops(!), And was it "Freestyle", the tablet based system that let C-Level execs annotate documents etc with writing and sound that also failed spectacularly because it was incorrectly positioned in the market and with price. Document Age Processing is gloriously absent, too. I used to work for Wang UK as marketing manager for the VS, PACE, Woffice etc in the late 80s, but haven't the competence today to write any of these up as articles or segments of this article.

What about the DVX, Digital Voice Exchange, that was Wang;s contributor to the UK government's DTI Office Automation Pilot programmes in the early 1980s?

There is also the Codd & Date "Hospital Case" 4GL data dictionary challenge that Oracle could only do with great difficulty, but with which Wang won regularly with PACE

Wang was a spectacular company with amazing products. It just ignored open systems and failed

Glad I didn't have to launch Wangcare in the UK, though!

Was the alleged advert "Wang, the chink in IBM's Armour" apocryphal?

Fiddle Faddle 00:04, 26 April 2006 (UTC)

Added some of this stuff today, some as sectstubs. Fiddle Faddle 21:41, 22 May 2006 (UTC)

"Minicomputer", "computer," or "mainframe"?

I think I have the ability to be authoritative here, only because I was employed in the UK to be the marketing manager for the Wang VS minicomputers. Since Wang defined them as minicomputers, and since the marketplace bought them as minicomputers, then it stands to reason that, minicomputers they were. All marketing literature, all press releases referred to them as minicomputers. Offerings against which they competed were minicomputers. Accordingly I have altered the page to reflect that. The discussion about what is and is not a minicomputer is interesting and acadeic. Nonetheless Wang defined what it sold as a minicomputer. Fiddle Faddle 06:16, 26 April 2006 (UTC)

Thank you very much. There's a difficult balance here between including legitimate information on new incarnations of the VS and allowing promotional language to creep in.
Your recollection and Dyl's both reflect my own recollection of how the VS was regarded and marketed, and my understanding of the word "mainframe." Like many terms used in computing on the marketing side there's no bright-line definition, so you should expect to see continued discussion here.
Because of Wikipedia's verifiability policy, and because all Wikipedia users are essentially anonymous, "authority" is mostly based on people's ability to convince others. What would be very, very helpful is if you have some actual Wang publications you could cite that called the VS a "minicomputer." Better yet, advertising in some print publication like Computerworld or Datamation.
I have a Wang golf umbrella, but that just has the logo on it! :) This is where recollection suddenly becomes hazy because it has been challenged. I can "see" the word Minicomputer in every description and every product launch piece of collateral I produced. I launched the VS 5000 at the Hippodrome as a minicomputer, and even have the autocue roll somewhere, sad man that I am. But I have nothing that I can produce to show the word "minicomputer" physically anywhere Fiddle Faddle 13:36, 26 April 2006 (UTC)
Ah, the VS 5000. My stint at Wang started midway through Fred's three-year turnaround plan, 1-3-5-10-20, and ended in the outplacement center along with about half of R&D in 1991. Dpbsmith (talk) 14:35, 26 April 2006 (UTC)
IMHO, In the article, the first appearance of the word "minicomputer" should be footnoted with a source citation showing that this was the term Wang itself used. It would be perfectly appropriate for the footnote to also mention that people selling modern VS successors call them "mainframes." This ties down the facts and also gives an objective rationale for using the term "minicomputer" in the body of the article. Dpbsmith (talk) 12:26, 26 April 2006 (UTC)
I've gone ahead to add such a note, but I have a question. Some quick Googling attempting to find Wang advertising of the 1990s turns up a few things like NYT stories obviously based on Wang press releases... I am getting the impression that Wang itself avoided using either the term "minicomputer" or "mainframe" and simply used the term "computer," as in "Wang VS computer" and "multiprocessor computer." Comments? Dpbsmith (talk) 12:48, 26 April 2006 (UTC)
Well, I propose that in describing the VS we should use whatever language Wang Laboratories customarily used during the time they were producing the VS, and that we should try to find out and reference what that was.
I'm beginning to think it was "computer," no more and no less. Dpbsmith (talk) 14:33, 26 April 2006 (UTC)

It gets harder the more one searches. On an intellectual level the VS 5 was never a mainframe nor could it be considered as such and the VS7000 series, based on the VS300 that was released a little before it worked properly (we have the smoke, we have the mirrors, let's go to market!), especially the bigger machines, could easily be considered to be mainframes.

When I launched the VS5000 in the UK it was positioned as something like "The smallest minicomputer in the world" Odd that we had about £120,000 to spend on the launch!

What is clear is that Wang never referred to its VS machines as mainframes. Equally clear is that its target market was against Data General, DEC, Prime Computer, Honeywell, and to an extent McDonnell Douglas, all of who were (surely?) undisputed minicomputer vendors of that period.

A google trawl brought very few people who referred to any "class" of computer, but who talked about their VS 100, or their VS System. One such is Voyager Systems who probably know what is what since they've been going since 1984 and were a Wang reseller back then.

The frustration is the difference between knowing something, and proving it. I shall, of course, stamp my foot ad say it was a minicomputer because that is what it was, but can I prove it?

There is a 1987 item here, though scroll down or search for "Wang" Additionally the first paragraph of this paper of data dictionary "stuff has no need to refer to a minicomputer, but it does.

Fiddle Faddle 18:10, 26 April 2006 (UTC)

Wang's VS family of minicomputers

WANG AIMING AT IBM, from Boston Globe Archives Published on February 21, 1985

"The new print ads promote Wang's VS family of minicomputers, comparing them directly with IBM's System 36."

http://nl.newsbank.com/nl-search/we/Archives?p_product=BG&p_theme=bg&p_action=search&p_maxdocs=200&p_text_search-0=wang%20AND%20aiming%20AND%20for%20AND%20IBM&s_dispstring=%22wang%20aiming%20for%20IBM%22%20AND%20date(all)&p_perpage=10&p_sort=YMD_date:D&xcal_useweights=no

Main search url: http://www.boston.com/news/special/archives/?p1=GoToStorySearch_AdvancedSearch

Perspective 18:55, 26 April 2006 (UTC)

I'm not trying to be difficult... and I belong in the "it was a minicomputer" camp myself... but, first, there's a difference between the Boston Globe calling them "minicomputers" (which tends to confirm that that's how the world saw them) and the interesting and stronger statement that Fiddle Faddle made, that Wang itself regarded them (and marketed them) as "minicomputers." Also, I found another Globe article, which reads in full:
February 21, 1985 The US Department of Health and Human Services has awarded the federal systems division of Wang Laboratories Inc. a contract valued at up to $14 million for the nationwide delivery, installation, and support of Wang VS computer systems. Wang was selected through a competitive procurement to furnish and support computer systems over the next five years. DHHS has placed an initial order for $4.4 million. Under terms of the contract, Wang will provide configurations consisting of Wang VS computers, terminals, peripheral devices, and associated Wang Systems Networking communications and software products for DHHS headquarters in Washington, and in DHHS field offices in 52 cities nationwide.
Here, the Globe is calling them "Wang VS computer systems." Or... since "Wang Aiming at IBM" sounds to me like the Globe speaking, while this item sounds to me like the Globe reproducing a Wang press release, perhaps it is Wang that is calling them "computer systems."
When I get home I'll trawl through my own paper archive of Wangiana. I don't think I'll find anything relevant though. Dpbsmith (talk) 19:35, 26 April 2006 (UTC)


What is this "minicomputer" compulsion really about?

There seems to be an unfortunate compulsion here in search of justification -- the compulsion to call the Wang VS a "minicomputer" instead of a "mainframe" in spite of an abundance of evidence of what the Wang VS really is, regardless of what it has been called. If Wang itself designed a line of multi-user computer systems around the IBM 360/370 instruction set and memory architecture, with robust I/O subsystems, but called it a "minicomputer," a "calculator," or a "rutabaga," would that make it a minicomputer, calculator or rutabaga, or would it really be what it is: a mainframe?

From Wang document 715-4652, dated 1992:

   The VS Instruction Set
   We say the instruction set is an IBM 360 superset because we have 
   added instructions to facilitate interactive access to the processor.
   With under 200 non-privileged instructions the VS could be classed as
   as a moderate instruction set (MISC) machine as contrasted to RISC
   (Reduced Instruction Set Computing) and CISC (Complex Instruction Set
   Computing).  The VS is so close architecturally to an IBM mainframe
   that one of the world's leading database suppliers uses a common set
   of object code for both architectures.

The key to this is "mainframe computing." Some of the attributes that may be used to distinguish mainframe computing from simpler forms of computing are:

Feature              IBM mainframes     Wang mainframes    Minicomputers

Hardware:
  DP-oriented instruc set  YES                YES               NO
  Decimal arithmetic       YES                YES               NO
  SVC instructions         YES                YES               NO
  Intelligent channels     YES                YES               NO
  Multiline telecomm       YES                YES               NO
  Synchronous telecomm     YES                YES               NO
  100+ disk drives         YES                YES               NO
  100+ tape drives         YES                YES               NO
  100+ printers            YES                YES               NO
  1000+ concurr. users     YES                YES               NO
Intelligent, microcode-loadable peripherals:
  Block mode workstations  YES                YES               NO
  Native workstations      YES                YES               NO
  Native printers          YES                YES               NO
  Native telecomm          YES                YES               NO
Memory & Addressing:
  Base/Displacement        YES                YES               NO
  Code/Static              YES                YES               NO
Networking:
  HDLC                     YES                YES               NO
  IBM SNA                  YES                YES               NO
  X.25                     YES                YES               NO
Clustering                 YES                YES              some
Virtual Machine OS         YES                YES              some
OS-level file systems:
  Record-based             YES                YES               NO
  Indexed                  YES                YES               NO
Job Queue                  YES                YES               NO
  Job routing by Class     YES                YES               NO
Print Queue                YES                YES               NO
  Print routing by Class   YES                YES               NO
Multiple native languages  YES                YES               NO
IBM BAL compatibility
  Basic Assembler Language YES                YES               NO
  BAL Macro facility       YES                YES               NO
  IBM-style linker         YES                YES               NO

There are some exceptions to the "NO" entries under "Minicomputers," but generally minicomputers were not business oriented although many were forced to business data processing, usually with the aid of significant third-party software and hardware. I worked with minicomputers from their inception with the PDP-1 in the 1960s through the PDP-11 and VAX in the 1970s. From machines that were "bare metal" with no software at all through machines that had real time OSs available, at no time did I ever have the luxury of readily available business data processing software, nor was it easy on any of them to write business software in machines that had no decimal arithmetic and in which it was awkward to handle records and fields characterstic of business data processing.

All computers were mainframes in the beginning, from ENIAC and others that occupied entire rooms and were measured in tons of weight. The IBM 360 is widely acknowleged to have been the first production mainframe of significance. The term "mainframe," though, did not come into use until the advent of the minicomputer, the first of which was the DEC PDP-1. Although the PDP-1 was large, consisting of multiple equipment bays and a full, 360-like console, it broke the mold by simpler architecture and instruction set and a price around $100,000. More or less from that point on, "minicomputer" came to mean leaner and less expensive than IBM's "big iron", which came to be referred to as "mainframes."

Minicomputers as a class were distinguished by smaller word sizes, binary-only arithmetic and, in their early days, the absence of instructions to support character fields and character operations. They were mostly word-oriented. Nor did early minicomputers come with anything that could be remotely described as business-oriented operating systems or languages. The PDP-1 was an 18-bit word machine with nothing more than an assembler and a few utilities. The PDP-8 was an 12-bit machine, the PDP-9 was 18 bits again, and the PDP-11 was a 16-bit, byte-addressed machine. CDC produced a 12-bit machine, the 160-A, although it was the size of an office desk. 1970, roughly the zenith of the minicomputers, saw in the vicinity of 300 offerings from companies as diverse as infant Data General, petroleum industry Schlumberger and aircraft manufacturer Boeing.

Minicomputers found use primarily in data acquisition, process control, universities, and message switching. They were specactularly unsuited for business data processing, having less ability to process business data than the 1950's punch card equipment that gave rise to mainframe data processing. Here is a very good description of what minicomputers were. It matches my experience.

Minicomputers

In contrast, mainframe computing has traditionally been oriented around business data processing, with decimal arithmetic at the hardware level and instruction set and architecture intended to efficiently run COBOL. A COBOL compiler on an IBM or Wang mainframe is virtually an assembler, as most COBOL verbs translate to single machine instructions and decimal storage representation of numbers is typical. And the resulting machine instructions in the IBM and Wang mainframes are essentially identical.

An incidental attribute of mainframes and supercomputers was that they were usually controlled by a smaller computer. The large VS models are controlled by a PC, the System Control Unit, that has a special interface giving it control over the VS, its I/O bus and all its co-processors. The PC loads the CPU with CP microcode, and loads all storage I/O Coprocessors with microcode for detection and presentation of IPL volumes. The PC can also run diagnostics on most types of I/O Co-processor, and keeps logs of power supply voltage exceptions. The small VS models have an onboard processor, the RCU, that performs many of the same functions. This is the equivalent of an IBM "service processor."

It is true that Wang generally avoided referring to the VS as either "mainframe" or "minicomputer." Later, some Wang materials referred to the VS as a "minicomputer." Later, when both terms had fallen into marketing disfavor, Wang began referring to the VS as a "server," the new term of popularity. One misinformed Wang salesman once lectured me about how the VS was primarly a WP platform, not a DP platform. It matters not what people called the VS, nor even what Wang called it. It matters what it was and is, which is illustrated in the foregoing table.

Wang's target market was always IBM, IBM, IBM. Others such as DEC were incidental. Dr. Wang seriously expected to overtake IBM. It was personal. IBM had screwed him over the magnetic core memory patents.

The marketplace did not buy the VS as a "minicomputer." The marketplace bought the VS as a less expensive alternative to IBM mainframes, and generally put the VS to the same work as IBM mainframes. Wang VS systems found use in all industries, often being the enterprise processor in businesses up to several thousand employees and $500 million in annual sales (turnover).

"On an intellectual level the VS 5 was never a mainframe..." On an intellectual level the Hercules IBM 360/370/390 emulator project people would disagree with you. They evidently define "mainframe" as a computing paradigm:

The prize for the world's smallest mainframe probably goes to Ivan Warren, who claims to have run VM/370 under Hercules on an iPAQ 5450 handheld PDA.

The VS5, like all the other models of VS, runs the same VS OS and the same applications as the others. It is simply a smaller machine. All system and application VS software is seamlessly compatible across the entire line spanning over ten generations of CPU and packaging hardware and close to 30 years. To say that the VS5 cannot be a mainframe is like saying that S/390 chipsets on a PC or RS/6000 I/O card cannot be a mainframe. In both cases the smaller machine doesn't have the large cabinets and robust I/O connectivity and throughput, but the computing paradigm remains -- business data processing in an IBM mainframe architecture, running the same software as the Big Iron. To claim that the large VS models are not mainframes is, well, just mistaken.

The physically largest Wang VS looked like this: VS12550 with room for two CPUs, eight memory boards, 15 I/O Coprocessors and a couple of square yards of cable connectivity on the backpanel. Each Coprocessor could handle 128 native workstation/printer/telecomm/etc devices, 14 SCSI devices, 32 async ports, 8 synchronous ports, an array of image jukeboxes, etc.

Right. It looked about as large as a PDP-1 or a VAX-11/780 "minicomputer" and much smaller than an IBM 7094 or IBM 360 or IBM 370 "mainframe." Dpbsmith (talk) 19:52, 22 May 2006 (UTC)
Sorry, I worked with the PDP-1 and the VAX-11/780, the latter of which was rack-mounted. The PDP-1 was later termed the first minicomputer because of its price, which in turn was a result of engineering a very simple computer with no business data processing capability at all. Its size had nothing to do with anything. The IBM 7094 shouldn't be in the discussion because it predated the canonical mainframe, the IBM S/360 and successor 370, some of whose models were smaller than some of the VS models. BUt the point of the picture was not just physical size, although it seemed necessary to address that. The pictured VS was capable of operating about an eighth of an acre of storage and other I/O equipment in a classic, raised-floor, air conditioned computer room, as well as hundreds of native workstations each up to 2,000 ft distant, and to interoperate with campus broadband networking called Wangnet and with numerous other VS systems in integrated networks. This is not a minicomputer.
You guys keep shifting between physical size, ignoring the large VSs when it suits you, and citing the terminology you used in referring to the VS, which is evidence of nothing. You insist on ignoring what the VS actually was, which was an improved IBM 360 with later updates to match 370 instructions. You insist on ignoring architecture and computing paradigm, which are what really define "mainframe" in its most common usages. There is no point in continuing this. I'll just go on making New VS systems and you can just go on enforcing misunderstandings of a system you no longer work with and apparently never worked with very closely. I've worked with most of the VS models from the VS80 through the VS18950 over the course of 22 years and contiue to do so. Since 1995 my contacts are with the VS group at Wang/Getronics, many of them going back to the original VS80, a couple of them now working for me -- people going back 26 years to the VS100, its microcode, the VS OS, and going forward through the course of the VS through the latest.
FYI, the VS5 and other small VS models were never Wang's bread and butter. That is to say, if that's what you think of when you picture the VS, you're missing what was really going on. The money was in the large systems and in extensive arrays of outboard devices. The money was in the multiple VS300 systems at Exxon Company USA with broadband Wangnet networking throughout the building, in the 45 VS systems operated by Mellon Bank and its subsidiary, Mellon Mortgage, in the clustered large VS systems at Kent Electronics with a network of smaller satellite VS systems in branch offices and manufacturing locations, in the vast network of VS systems operated by the U.S. State Department, in the 80 or so large VS systems operated by The Hartford, in the still-operating and upgraded seven VS systems at Ark. Dept. of Health, the largest fiber cluster of VSs in the world, serving something like 2,200 users 24 x 7, in the still-operating VS systems that run the Colorado Lottery, and on and on. The money overseas was in large, multi-VS systems like the port facilities management company in Europe, the huge number of VS systems which ran Nor-Cargo of Norway, the banks that depended on VS systems, like the New Zealand bank that uses multiple VSs today, among other things to do LU 6.2 with other banks for interbank transfers, in the large Deutschebank VS facilities, the Seoul Police Department system of large VSs that drove the real time wall projection of vehicles and incident sites at police headquarters, et al. These were multimillion dollar accounts with annual support and maintenance in six and seven figures. Wang viewed the small machines as things they just had to offer to remain credible and viable. And you want to characterize the entire VS line by the small VS models? It makes no sense unless that happens to be the limit of your experience with the VS. Or unless you harbor something against Wang. If the latter is the case you should be advised that none of the people who laid you off or otherwise annoyed you are still there. Tjunker 03:07, 30 May 2006 (UTC)

BTW, the 2,000-user/device VS OS, 7.54.20, is in final test for release. In time the configuration limits, formerly based on 15 physical I/O slots, will probably expand, since I/O Co-processors are now virtual and inherently unlimited.

Tjunker 13:29, 22 May 2006 (UTC)

You ask: "What is this 'minicomputer' compulsion about?" I answer: it's nothing more than pushback against your pushing your "it's a mainframe" point of view, which, as far as I can tell, has raised the eyebrows of everyone else who has cared to discuss the matter here.
Minicomputer, mainframe, etc. are not and never have been engineering terms with well-defined meanings; they're cultural and marketing terms and several of us concur in agreeing that not only did the press call Wang's products "minicomputers," but that Wang itself (including Dr. Wang) seems to have avoided the term mainframe. In "Lessons" he clearly contrasts Wang's machines with IBM's "mainframes."
The contrast to be made was between the usability of the VS and the difficulty of using the IBM mainframes, hobbled as the latter were by dragging boatloads of batch legacy forward in their operating systems. The innovation of the VS was that it was designed from the ground up to be an interactive system as well as a mainframe-class batch system. Generally, interactive systems can run batch processes simply by running them from workstations, but batch systems implement interactivity only with great difficulty and the tooth-bending sound of great square pegs being pounded into round holes. CICS was to the batch IBM OS what Windows 95 was to MS-DOS -- a layer added as an afterthought to try to adapt the underlying OS to something it wasn't. Wang implemented true interactive COBOL and transaction commit/rollback quite easily because interactivity and soft recoverable file capability were built into the OS. Tjunker 06:05, 30 May 2006 (UTC)
It's very much like Sports car. You can argue as much as you like about the definition—because there isn't one. You can say objectively, based on numerical characteristics, whether some car "is" a Formula One car or an Indy car. And you can say objectively whether the maker of a car calls it a sports car. But you can't say objectively whether a car "is" a sports car.
The IT community called Wang's offerings "minicomputers." You give cogent reasons as to why that might have been unfair, but that's what they were called. To call them "mainframes" is misleading enough to be a problem. It would be like referring to the Xerox Alto as a "PC." You can make out an excellent case for the Alto having many of the characteristics of a modern PC, and Xerox even called it a "personal computer," yet calling it a PC would be inaccurate.
I feel that you cannot be regarded as neutral on this, because you have a direct personal interest in putting the VS's best foot forward.
No one militating for truth and reason is neutral. Neutrality in this context seems to mean complacency with an inaccurate, skewed description of the VS. Why should I be comfortable with that? Ever since I became concerned with Wang allowing the VS market to slip away -- the market which supported me -- I have been trying to clarify and provide information about the VS. Since the VS now has a new lease life by being recast in new hardware, it is even more important to clear up issues such as what the VS really was and is. Since 1995 my [[Unofficial Wang VS Information Center has been providing the best VS information I have been able to assemble, at no cost to anyone or anything listed. User:Tjunker|Tjunker]] 06:11, 22 May 2006 (UTC)
However, if you have any source citations that show that Wang or anyone else (other than the company you're presently connected with) used phrases such as "Wang VS mainframes?" That would be far convincing to me than tables of data. I've thrown out all my copies of "Access to Wang" magazine, but perhaps you have some? What terminology did it use? Dpbsmith (talk) 14:36, 22 May 2006 (UTC)
P. S. On looking at the current state of the article, I think that it might be appropriate to avoid the contested terminology altogether, refer to the VS as the "Wang VS Computer" throughout, and move all the discussion of terminology into footnotes--but without losing any of the references. Dpbsmith (talk) 14:38, 22 May 2006 (UTC)
"it's nothing more than pushback against your pushing your "it's a mainframe" point of view"
Thanks for the frank admission that there's nothing to it except pushback. I've been pushing for an accurate classification of the Wang VS and you've been pushing back for no reason other than preconception and resistance to reality. Perhaps you can refer me to the Wikipedia reference page where obstinate pushback is listed as legitimate objectivity. Tjunker 03:23, 30 May 2006 (UTC)
"To call them 'mainframes' is misleading enough to be a problem. It would be like referring to the Xerox Alto as a 'PC.'"
IMO that is grossly inaccurate. The very first Wang VS was more capable than any of the IBM 360 models and comparable at least to mid-range IBM 370 models. You use inappropriate technical analogies but the only evidence you present is that somebody called the VS a minicomputer here or there. How about trying to show how the VS was like a PDP-8 or PDP-11 or a Varian 620 or any of the other classic minicomputers? The fact is that the VS has nothing in common with any of the minicomputers of the 1970s or 80s and everything in common with the IBM 360/370 family. Tjunker 00:44, 30 May 2006 (UTC)
Let's get even blunter.
Sure, let's get more blunt: You're concerned with what you and others may have called the Wang VS; I'm concerned with what it was and is. You've presented no evidence other than references and claims of what it may hav been called by some; I have provided abundant evidence of what the Wang VS was and is, the close similarity to IBM mainframes of the era, and the contrast with what minicomputers were in the time of the VS. I lived through the minicomputer era. I know what minicomputers were in the 1970s and 80s. No way is the Wang VS, built along the lines of the IBM 360/370, a minicomputer in any way, shape or form. Tjunker 00:59, 30 May 2006 (UTC)


I was first employed by Wang to be a support analyst in the city of London for its minicomputers. I was promoted(!) to be the UK Marketing Manager for, among other stuff, its minicomputers.
We knew them as minicomputers
You could have known them as "analog handheld calculators" but that wouldn't have made it so, either. Tjunker 03:26, 30 May 2006 (UTC)
Now let's apply the duck test. Whatever the instruction set inside them, even if Dr Wang himself were playing with the beads on the little abacus that we know is at the heart of every Wang VS, it looks like a minicomputer, smells like a minicomputer, was known by Wang as a minicomputer (with or without the marketing puffery of "Super"), and it quacks like a minicomputer. Thus it is, guess what? Yes, a minicomputer.
"Whatever the instruction set inside them..."
The combination of IBM instruction set, control store and microprogramming, 24-bit base/displacement byte-addressed memory, machine architecture, compatible models, decimal arithmetic, independent I/O processors, etc., is exactly why the VS is a mainframe. Those are among the things that distinguished the IBM 360, as described in System/360 and Beyond. Tjunker 01:16, 30 May 2006 (UTC)
"...it looks like a minicomputer..."
You might actually read the definitive Minicomputers reference and others I provided above. At 600-1000 pounds in desk-size and 5'-tall cabinets, the mainline VS models look nothing like any minicomputer I ever worked with or researched in the 1970s and 80s. No VS was ever rack mountable. Large VS sites consisted of one or more VS system units with rows of disk storage and tape cabinets behind them. If the only VS systems you ever worked with were small format ones it would be understandable that you might mistake them for minicomputers. The small models of IBM 360 were also sometimes called "minicomputers" but that didn't change the fact that the 360 was the family of models that established the paradigm that was later called "mainframe" to distinguish it from smaller, mostly non-DP, mostly incompatible systems, nor the fact that the Wang VS was designed in the same paradigm as the IBM 360 and shares most of the 360/370 characteristics. Tjunker 01:25, 30 May 2006 (UTC)
"...smells like a minicomputer..."
What is that supposed to mean? I think it means simply that you think of the VS as a minicomputer. That's not evidence of anything. Have you worked with the VS300, 7380, 12550, 12650, 16850, 18950? Have you worked with a VS that handles 300-500 users? Have you worked with clustered VS systems? Systems with 50-100 disk drives, 10-20 tape drives, hundreds of printers, rooms full of laserdisc image jukeboxes, with Wangnet, with IBM 360/370 architecture and instruction set, with machine and OS support for data processing languages and file types? If you examine the Wikipedia mainframe page, it pretty well describes the Wang VS. I think you must have been exposed only to a small subset of the VS product line else you would not have been laboring under this misapprehension. Tjunker 01:33, 30 May 2006 (UTC)
"...was known by Wang as a minicomputer (with or without the marketing puffery of 'Super')..."
No, the marketing puffery was what you describe as the U.K. practice of commonly referring the VS as a "minicomputer" and the later adoption by U.S. home office of "server" to keep up with IT fashion trends in the mid-to-late 1990s. It seems likely that Wang's reluctance to classify the VS early on had more to do with marketing considerations than anything technical. There may have been reasons early on to avoid appearing to go head to head with IBM at the high end, then later, "mainframe" became a pejorative, then both "minicomputer" and "mainframe" were pejoratives and "server" was the fashionable term. Both the timing and the nature of the marketing considerations of terminology may have been quite different in various national and regional markets. Then, too, Wang's control over their network of subsidiaries was never ironclad. The U.K. could probably have called the VS a "sconecomputer" and gotten away with it. Closer to home a Houston VS sales dweeb once told me the VS was really a WP machine and only did DP poorly, but that bizarre reversed view didn't make it so. Tjunker 02:39, 30 May 2006 (UTC)
And the Wang User Group in the UK talked about...... minicomputers. I know, coz I was there doing it.
Apparently you guys thought of the VS that way. You could as easily have called them "personal computers" and it would be just as relevant -- it has nothing to do with what the things were. Tjunker 02:41, 30 May 2006 (UTC)
I don't really care what Getronics has nmanufactured. That can be a Supderdociousframe if it likes. But the Wang VS machines were minicomputers.
Have you really managed to miss the whole point? First, Getronics was later in the game that you seem to think. They acquired Wang Global in late 1999, after the release of the VS18950 CPU. After that all Getronics did was allow the VS group to finish what they had been doing -- engineering and releasing the last two VS CPU boards -- the small-format VS6760 and VS6780 based on the VS18950's CPU chip and providing software support. Those last two new CPU upgrades were released in early 2000 by GetronicsWang, a name stop on the way to eradicating the Wang name everywhere except on manuals and machinery. The Superwhatever you seem to be thinking about must have been manufactured by Wang, who designed and manufactured all the VS models from the 1977 VS80 through the 199 VS18950. No Getronics there at all. All the VS models had remarkable compatibility, the same design principle that set the IBM 360 apart from previous computers. All run the same OS and applications, and were all based on the IBM 360/370 with enhancements. All the large VS models are IBM 370-class machines, not minicomputers. The small VS models could be termed minicomputers only in the sense that small IBM 360 models were sometimes termed minicomputers. But all VS models were far and away more robust and capable than the early 360 models. Look both of them up. Tjunker 02:49, 30 May 2006 (UTC)
I suppose we could start the idea that they were actually very fast microcomputers... But I'd resist that as strongly. This is as silly as suggesting that Michael Jackson is caucasian.
Fiddle Faddle 16:32, 22 May 2006 (UTC)
For any who may be interested, here is a rather full large VS configuration. It has over 3.3 TB of regular disk, 147 disk devices altogether, some of which are indeterminate RAID (it self-configures and doesn't have to be specified by size in the config), 150 native workstations, 300 networked workstations that could be remote VS sessions or Lightspeed NVS desktop PCs, 42 2000 lpm band printers, 129 synchronous lines that could be other VSs or SNA lines, 17 tape drives, mostly DLT and 4mm, 4 image devices and is part of a 16-VS fiber cluster. A total of 805 devices is configured, some of which are auxiliary devices that support primary ones, such as extra windows for workstations. All the configured devices could be connected using only Wang equipment. Other than Lightspeed NVS for TCP/IP desktop PCs, all the equipkment and interconnects and cables and cabinets are Wang equipment. It's an extreme configuration, especially for the VS738K shown, but the same could be configured on any of the large-format, 15-slot VS models, and certain additional IOCs could be configured in an expansion cabinet. This is decidedly not minicomputer territory, and far exceeds anything the IBM S/360 could do, although it might not reach to the limits of the S/370's high end. Tjunker 08:28, 30 May 2006 (UTC)
An amusing rebuttal, well worded, well thought out and painstaking. Regrettably also not borne out by the facts of the market at the time. Or, to put it another way, wrong.
A big VAX was large and heavy, so were Tandems, Primes and so many others. But they were also, in the parlance of the day, minicomputers
The distinction is genuinely irrelevant save for one thing. They were marketed as minicomputers. Thus it is important in the context of the marketing.
Does the term get used today? Well no. We speak in other terminology. It only matters, so far as it matters at all, in the way Wang chose to market its wares. The VS was marketed and positioned against minicomputer vendors. Ergo, whatever you think it was, whatever you attempt to prove it was, it was a minicomputer.
After all, who in their right mind as a CIO (a term not used then, of course!) would buy a Wang VS as a corporate mainframe?
Fiddle Faddle 09:23, 30 May 2006 (UTC)
if you asked me what's this, I would unhesitatingly say "a minicomputer." Mainframes look like this.
But, really... anybody dispute that they were very frequently referred to as "Wang VS computers?" If we're all agreed on that, is there any reason why we can't just call them that throughout the article (at least when not discussing issues of Wang's product positioning and marketing strategies)? Dpbsmith (talk) 13:20, 30 May 2006 (UTC)
I thought the first picture was of a small freetanding wardrobe, and the second a series of wardrobes. So they are computers, then? Now I know what I was employed to do. It would be far more productive, I think, if we actually embellished the article. There is so much missing stuff. The Document Image Processing was a first, at least at this level. Freestyle was amazing and destined to fail. The DVX cursed us all with voicemail! WP and email removed at a stroke the job of "Secretary" as WP alone rmeoved "Typist". Fiddle Faddle 13:37, 30 May 2006 (UTC)

Tempest Proof Equipment

Getting citations for this lot may be hard because it was all secret squirrel stuff back then.

Wang had long tempest proofed the OIS and it was reputedly used at the trial of Oliver North to process the court papers.

It also had quite an amazing range of tempest proof properly compatible PCs, with an extra chamber for placing extra cards. They were only marginally larger than the then range of PCs and a lot heavier because of the shielding.

I can't recall if they ever tempested a VS, but Wang was a major US Government contractor for highly secure equipment.

Fiddle Faddle 17:42, 26 April 2006 (UTC)

That Wang was producing TEMPEST-certified equipment was surely no secret. Dpbsmith (talk) 19:37, 26 April 2006 (UTC)
NEW WANG COMPUTERS
Boston Globe
March 8, 1989
LOWELL -- Wang Laboratories Inc. said it introduced its Tempest Wang Integrated Image Systems, which combine new security measures with imaging technology. Wang also said it introduced a high-performance minicomputer and personal computer systems with Tempest security. Wang said the WIIS-T enables users in sensitive information-processing environments to capture, store, retrieve and communicate millions of image pages currently stored on paper. Users can access, display, manage and manipulate images, text and data on a single workstation. System prices range from $150,000 to more than $1 million.
Section: BUSINESS
Page: 46
That had to have included a VS. Dpbsmith (talk) 19:40, 26 April 2006 (UTC)
WANG INTRODUCES LOW-COST, 32-BIT MINIS
Boston Globe
April 8, 1987
LOWELL -- Wang Laboratories Inc. added two low-cost 32-bit minicomputers yesterday to its TEMPEST systems product line, used for security-oriented data processing. The entry-level VS 5T and VS 6T, priced at $13,500 and $21,500 respectively, are compatible with all VS systems, Wang's main line of super minicomputers, the company said.
Section: BUSINESS
There ya go. A TEMPEST-certified VS and "minicomputer" together in one sentence. Of course, they're "super minicomputers." Much better than "feeble mainframe," methinks. Dpbsmith (talk) 19:43, 26 April 2006 (UTC)
Odd, isn't it. We were all told "not to mention it" ROFLMAO Fiddle Faddle 20:44, 26 April 2006 (UTC)

Trivia

WangCare

WangCare, the service offering that was meant to be launched globally, was not ever likely to be a success in the United Kingdom. The similiarity to Wanker meant that the UK team refused to use the word at all, despite explanations to the USA of the meaning. The term "wanker" does not travel well beyond the old "British Empire".

Fiddle Faddle 19:43, 17 May 2006 (UTC)

Adverts

  • The man in a football helmet on a trolley, destroying the office appeared on US TV, as did the "Electric baby". Do these adverts prove the insanity of Wang or the ad agency? Fiddle Faddle 19:43, 17 May 2006 (UTC)
I don't remember those. The ones I do remember were the "jargon-talk" ads that aired on radio about 1988. They were incredibly irritating to me and to most listeners. They featured what sounded like Wang salespeople throwing around acronyms and initialisms and displaying a perceptible contempt for their prospects. One of them involved a salesperson demonstrating something about interoperability or networking. The prospect found it so amazing that he couldn't believe it was an authentic demonstration and "would you believe it, he checked the wires!" Dpbsmith (talk) 20:23, 17 May 2006 (UTC)
Ah. Here we go. Boston Globe, April 21, 1987: "Wang's New Pitch: Jargon" "Wang Laboratories Inc. is letting the computer jargon flow freely in a new advertising campaign that lets viewers eavesdrop as actors posing as Wang salespeople talk shop.... the campaign has prompted competitors to ask privately whether the ads are unintentionally telling laymen they need a college degree to operate a computer system.... Jane Carpenter, Wang's ad manager, said "computer-specific terms" may be just what is needed to help the company's ads stand out. 'We wanted the ads to be intrusive.'" The story quotes one of the ads:
Let me tell you a story about connectivity and networking that I think you'll get a kick out of," a saleswoman tells a colleague in one commercial. Before you can loosen your tie, the story is going like this: "We take a DEC workstation and via a Wang PBX we get it talking to his own IBM mainframe through our Wang VS," she said. "And, oh yeah," she pushes on, "the DEC workstation is talking to an IBM PC on the other end of the room via Wang Office."
I think that's the one that ends and he checks the wires! as the salespeople chortle merrily.
I'm sure I remember "those annoying Wang ads" becoming a bit of a by-word in the media for a while during and after the campaign... "so and so uses so much jargon he sounds like one of those Wang ads," etc. Dpbsmith (talk) 20:34, 17 May 2006 (UTC)
I never saw it myself, but I heard that at one Wang user's meeting, after the VS 5000 had been announced, people were handing out T-shirts with pictures of VS 5000 and the caption "It's not how big your Wang is, it's what you do with it."
Oh yeah, I remember a Wang calculator print ad... at that particular time, that particular model of calculator had more storage than the competitors... and the ad said something about "Wangs for the memories." Dpbsmith (talk) 20:38, 17 May 2006 (UTC)

Unix

With the best will in the world, how on earth can one get a citation for something as intellectually sound as the two items that are flagged here for a citation? Fiddle Faddle 21:58, 23 June 2006 (UTC)

  • I didn't put the tags on... but, Fiddle, with the best will in the world, the verifiability policy says that everything in Wikipedia must be traceable to a published, reliable source. Even though it is often honored in the breach, that's the policy. I don't have time right now to see whether Riding the Runaway Horse has anything to say about this. In my personal opinion, a source citation to something like Access to Wang would do, too. Wikipedia has a strict policy against original research. Ultimately, if these times can't be sourced, they should probably be removed. However, it is usually the case that when "everyone knows" something, someone will have written about it. Dpbsmith (talk) 23:59, 23 June 2006 (UTC)
    • I don't dispute or disagree with the policy. Never have and never will. It is just not (easily) citable because there will be no hard facts there. There will, if course, be opinions. The issue is not as simple as it could be had the company never made PCs. But it failed, totally, to enter the Unix market while pretending it might just possibly have a valid implementation on the VS as an emulation. The players it competed against included the nascent Sun, SGI (yes I know they were really renowned as graphics engines) and other hot, fast Unix boxes because the software market moved into Unix and Open Systsems. And Wang could not run them. But those of us at the sharp end knew it. If I could track down the Unix Product Manager we could start to verify it, but I have so far failed to do so. Fiddle Faddle 06:58, 24 June 2006 (UTC)

More erroneous statements creep in

"However, the market for these mainframes was ultimately conquered by PCs."

That has to be one of the dumbest errors that has been introduced here. No mainframe market, Wang VS or other, has ever been "conquered by PCs." Most Wang VS systems of any size and substance that have been replaced have been replaced by large systems running either a major COBOL dialect or major RDBMS.

The host horrific crash and burn of an attempt to convert a company off the Wang VS happened to Kent Electronics of Houston, TX. They hired Sun Microsystems to get them onto Sun Solaris, using boxes substantially more capable than "PCs." It crashed and burned in the final live load testing. None of the involved contractors and subcontractors was able to get it working... not Sun, not Dev-Tech (vendor of the WISP conversion tool), not the COBOL vendor and not the third-party indexed file system vendor. Kent kicked Sun out, demanding that Sun buy them another large VS on its way out. That was reportedly done, and Kent added the system to its existing two, making a 3-VS fiber cluster serving about 700 concurrent users. Kent later did an outsourced Indian conversion to Informix on very large AS/400.

It's evident that some of the contributors here who claim to have worked in the VS field never worked in the U.S. with anything more than tiny systems. A "typical" Wang VS site in the U.S. was a small to medium business running the enterprise on the VS. WP was popular with a subset of the VS community but was never what drove the VS market. "Typical" VS application suites are in COBOL and run to 1-3 million lines of code. Larger sites almost always have and use the PACE 4GL/RDBMS, which supports HLI in COBOL and RPGII. PACE database files are also accessible from standalone COBOL, BASIC, RPGII, C, PL/I, etc. I have had many VS sites as clients but have never had one where the VS was primarily used for WP, nor any where the VS was not the enterprise system or at least ran a major division of the enterprise. I have never had a VS client who could consider replacing the VS with PCs unless he was ready to check into the rubber room.

To those of you who like to mention the VS5, VS6, VS5000: that was not the VS world. Those were tiny machines, although 100% software compatible with the larger ones. The VS world was characterized, at least in the U.S. and continental Europe, by large machines — VS300, 7000, 8000, 9000, 10000, 12000, 16000 and 18000 — and hundreds of users, often wide ranging WSN networks, high-volume printing, dozens of disk drives, tape drives of many kinds, etc. Anyone who was only exposed to the tiny VS models can't possibly have any perspective on the larger picture of the VS community and the kinds of uses to which the VS was put.

UNIX ran on the VS, but as an emulation running under the VSOS, and thus had major performance challenges.

The emulation part is just plain false. IN/ix, the Wang Unix product, was a port of one of the AT&T Unix versions. It ran natively as VS code, as a guest OS under VS/VM, the VS Virtual Machine OS, which had very low overhead. Native. No emulation.

One of the little known facts about the VS, though, was that its processors just weren't very fast. In fact the functionality and popularity of the VS was a testimonial to the efficiency of its OS and operating environment, an efficiency that allowed the VS to provide substantial services with what were in reality underpowered processors. The first VS, the 1997-8 VS80, came with 256KB of main memory and maxed out at 512 KB, but served up to 32 concurrent users. The fastest VS, the VS18950 released in 1999, has a clock speed of 58 MHz and supports 500-1000 users and hundreds of I/O devices.

TJunker 12:38, 13 October 2006 (UTC)

You said "To those of you who like to mention the VS5, VS6, VS5000: that was not the VS world. Those were tiny machines, although 100% software compatible with the larger ones. The VS world was characterized, at least in the U.S. and continental Europe, by large machines — VS300, 7000, 8000, 9000, 10000, 12000, 16000 and 18000 — and hundreds of users, often wide ranging WSN networks, high-volume printing, dozens of disk drives, tape drives of many kinds, etc. Anyone who was only exposed to the tiny VS models can't possibly have any perspective on the larger picture of the VS community and the kinds of uses to which the VS was put.
The VS world was characterised by machines of all sizes in all configurations, often with departmental machines networked to some sort of corporate hub. Just because you don't like a baby VS that could be installed by a secretary (I know, because I tested it with our Wang UK Marketing Department secretary who did it in no time flat from the box to a running machine) does not mean you can make it less valid.
The VS started out as heavyweight and frankly pretty low powered boxes. It ended up as a range of boxes of varying power. It doesn't need airconditioning and computer rooms to be a VS.
They were all very well implemented. Processors with efficient code meant Wang could stay in TTL logic well past the date competitors moved to ECL. Fiddle Faddle 14:24, 13 October 2006 (UTC)
For what it's worth, the last VS I personally happened to see in actual commercial use was a VS 6, at an appliance store named Poirier Sales and Service in Norwood, Massachusetts. I believe it was in the late nineties. (I'm not sure exactly how big Poirier is or exactly how you measure business size, but in the late nineties it had a single location, with about the same amount of floor space as, say, the appliance section of a single Sears store).
When I worked at Wang, I constantly noticed a disconnect between Wang's real customer base and what Wang wanted to be its customer base. Wang wanted to be a big player, and career advancement, certainly for sales and perhaps to some extent for everyone else, went to those associated with the big machines. So, to hear people talk about it, you'd think it was beneath the salespeople's dignity to call on anyone below the president of MellonBank.
So, within Wang circa 1990 anyway, I always felt there was a systematic self-deception that led people to think that the "typical" Wang installation was what Wang wanted to be a typical Wang installation, instead of what the typical Wang installation really was.
And, frankly, I think at least part of Wang's downfall was due to unwillingness to show loyalty to smaller customers, abandoning e.g. the people who bought OIS systems. At one point I heard some big pooh-bah market strategizer say openly that he wanted Wang's word processing software for PCs to fail because they wanted people to forget Wang word processing, so if the product succeeded it would remind people "the old Wang" and interfere with "the message" they were trying to send that Wang didn't do rinky-dink departmental computers any more. I don't know what they teach in b-school but I've always had a strong suspicion that trying to make products fail (or to deliberately "kill" well-liked products that are making a profit) is not a good plan. Dpbsmith (talk) 15:37, 13 October 2006 (UTC)

The latest error introduced was dpbsmith's edit on 8/19/08 of my addition about the announcement of end of support for the legacy VS. He altered my text to falsely state that existing support contracts would be honored via the New VS. Oh, if only it were so! My original text correctly stated that existing support contracts for the legacy Wang VS will be honored by Getronics (now Compcom) through the expirations of those contracts. The New VS has nothing whatsoever to do with support contracts for legacy systems. I will do my best to fix it without reverting to the original text, where the factual performance information was deemed to be excessive and promotional. Apparently the prevailing thought here is to give short shrift to facts. TJunker (talk) 13:48, 22 September 2008 (UTC)

My apologies. Please do correct the passage. I was trying to produce a short but accurate summary of what you had posted, and apparently I misunderstood your material. Dpbsmith (talk) 14:01, 22 September 2008 (UTC)

OIS Section

Needs complete rewrite. Poorly written and full of POV stuff. Thought seriously about deleting, but there really does need to be an OIS/Alliance section in the article. Rhsatrhs 04:37, 8 November 2007 (UTC)

POV stuff? Writing style is writing style, and that can always be improved, but it seems factual enough, albeit uncited. What am I missing that you can see here? Fiddle Faddle 08:32, 8 November 2007 (UTC)
"Secretaries commonly"... The word "commonly" is POV. Or at best, it is too vague. A more neutral statement would be that a design goal of the Glossary system was to enable typical users, most often secretarial workers in the workplaces of the time, to accomplish simple programming tasks.
"Oddly, the OIS overlapped the Wang VS"... The word "oddly" is clearly POV. Also, the VS came after the OIS, so if anything is odd it's that the VS overlapped the OIS, not vice versa. (But it is not odd at all. There were sound business strategy reasons for the overlap.)
"Familiar features"... The word "familiar" is somewhat POV. Familiar to whom? To everyone? Probably not. Familiar to people who installed and supported the two systems, sure.
"probably because WP did not figure into the original design of the VS but was added later"... The word "probably" is POV unless you have citations of conflicting opinions and can show that the majority point to this explanation.
"The most significant enhancement was"... The phrase "most significant" is POV. It is undoubtedly true, and I'm sure there are thousands of similar wordings on wikipedia pages, but there should be a citation of an authoritative source for the opinion, or it should be stated in a more objective way.
Rhsatrhs 12:42, 8 November 2007 (UTC)
I can accept the argument that the language needs making more formal. I was thinking you had spotted a huge point of view issue in the section, and just could not spot it. I think a mild edit will solve those issues you point out. Fiddle Faddle 15:30, 8 November 2007 (UTC)

The original Wang PC

"The original Wang PC was released to counter the IBM PC which had gained wide acceptance in the market for which Wang traditionally positioned the OIS system."

Misleading. During my time at the Wang National Technical Support Workshop in Sydney, Australia in 1979/80 I worked on a model that was designated as a 2200PCS, for Personal Computer System. This model was released in 1976 and thus predates the delivery of the IBM PC by many years. It was a self-contained unit with keyboard integrated into the main monitor case. It was internally designated as a 2200E or F model, depending on whether it used cassette drives or the newer 5" floppies invented for it by Shugart. Ref: http://www.wang2200.org/systems.html Haiqu (talk) 10:11, 23 February 2009 (UTC)

Do you think someone should tell this guy? — Preceding unsigned comment added by 136.154.22.22 (talk) 03:40, 3 August 2011 (UTC)

Don't Forget the WangWriter

This article is definitely a stroll down memory lane, as my first programming job was on the 2200 series in 1978.

One product missing from the Wang discussion was the WangWriter, which was a small CPM-based dedicated word processor that appeared sometime around 1980-1981 (perhaps 1981-1982; memory is a bit fuzzy). I believe the integral printer was a daisy-wheel printer for high quality output in an era where PC's (if they even had printers) used dot matrix technology.

The WangWriter was intended to run the same microcode as the OIS terminals and had a desk-side CPU/printer integrated into a cart, and a separate CRT and keyboard. As I recall, it was designed to help Wang protect the low end of its word processing business against emerging PC-based systems, such as Samna Corp.'s Ami Pro (don't remember exactly when that appeared, but it was far more sophisticated than the WordStar application that migrated to the PC from CP/M systems). Ami Pro was a particular threat since it grabbed a bunch of the ideas for its user interface from Wang and was pretty up front about that in its marketing pitches.

What made the WangWriter completely DOA was its price -- as I recall, it was about $10,000 versus a PC hardware cost of about $3,000 plus another $1,000 for a daisy wheel printer. The competitive market for word processing in the early days of the PC (WordStar, Ami Pro, Word Perfect) made the features of those systems evolve very rapidly compared to Wang, who was all about protecting the existing OIS business, where the user interface had been relatively unchanging since 1976 or thereabouts. I seem to recall that Wang sold only a couple hundred WangWriters and took a massive financial bath on the R&D and manufacturing investment. Even Wang sales people made fun of these things, though I don't recall whether it was due to the product itself or due to worry about what it would do to their average deal sizes, and thus to their commission payments.

So I think the WangWriter deserves mention because it is another example of the "too little too late" mode that the Company was in, where it didn't respond to major changes in the marketplace until after its doom was sealed.

I have some of my old 2200 stuff, but I'm sorry that I don't have enough verifiable information such as old manuals, price lists, etc. on the WangWriter to create a section on it myself...

Brent (talk) 22:04, 22 September 2011 (UTC)

WangWriter came along in 1980, about the same time as IBM's DisplayWriter. IIRC, it originally cost $7,500 (about $400 less than the DisplayWriter to be highly competitive with that; at the time, IBM's PC was still a year away, and the only "serious" PC competitors were the Apple II+ and the TRS-80).

The first incarnation of WangWriter had the original "Ergo" modular workstation (separate screen/electronics and keyboard) and a system unit that contained a stripped-down OIS master, 5.25" diskette drive and loud-as-a-jackhammer integrated daisy printer on the top. In mid-1981, WangWriter II came along with a smaller system unit (ditched the integrated printer and shrank to only about 12" wide), and support for up to 2 external printers (standard 928 connections).

The diskette form factor was incompatible with other Wang systems so unless it was connected via WangNet or WISE (or later dial-up TC) it was an island unto itself. (I tested a 5536 emulation program that made it appear as an archiving workstation if connected to an OIS-140; That was in mid-1982. I don't remember if it was actually released.)

A CP/M option came out sometime in late 1981 that ran on both the original WangWriter and the WangWriter II; however, as was typical of the era, CP/M and the native Wang sides were dual-boot and not compatible with each other (though you could exchange plain text files via diskette).

WangWriter's OS was a custom-written, single-user that mimicked OIS DOS (supposedly a subset based on OIS DOS but whether it was actually based on OIS or brewed just for the WW depended on whom I spoke with...it was like no one wanted to talk about it much). WP was a custom build that looked and felt like WPS or OISWP but with some seriously annoying quirks and inconsistencies. WP had trouble keeping up with typists who were better than 60wpm; early versions required conversion between WW and OIS document formats; documents longer than 30 pages almost always had to be recovered for no apparent reason; mixing portrait and landscape pages in the same doc had a nasty tendency to make it freeze; and OIS glossaries couldn't be used on WW if they used advanced programming features. Rounding out the core system was a custom, stripped-down List Processing workalike (IIRC, it was limited to something like 8K or 16K records per list; a reasonable limitation for mailmerge but meant you had to get the CP/M option and dBASE to use the machine for DP-like tasks...and export merge data via diskette to read-in on the Wang side). Overall performance was like a diskette-based WPS-5, not something in the OIS-140 or VS90 age. We did a 6-month pilot with 4 of them at HBO in 1982 and decided to stick with straight OIS for WP/light DP purposes.

I have never understood why people insisted that programs like MultiMate or Samna (1984) were Wang WP "clones." They bore little resemblance on the UI, and other than mimicking Wang's "Copy|Move What?/highlight/To Where?" they were every bit as clunky and cumbersome as WordStar or XYwrite, or WordPerfect. The only PC WP package that I ever considered a true Wang clone was Leading Edge's word processor. I was so disappointed that LEWP wasn't marketed better...I vastly preferred it to WordPerfect.

WangWriter was a nice idea but, in typical Wang fashion, they just couldn't figure out how to articulate and sell its abilities. (That inability on their part was how my business got started, and gave me quite the revenue stream from developing applications for the OIS and helping sell the hardware in the NYC market for much of the 1980s.)

Jeff (talk) 16:14, 7 June 2012 (UTC)

Wang 3300 -- missing machine

I am not a wikipedia expert, but I'd be happy to work with someone fluent in the ways of wikipedia to create a section on the Wang 3300. This machine is missing from the Wang Labs entry, and should have at least a nod. It predated the Wang 2200 and though a commercial failure, taught Wang a lot and lead to the more successful 2200 family.

The 3300 was intended to be a general purpose computer ala the PDP-8, but Wang missed the mark. The BASIC interpreter for the 2200, although a complete rewrite, inherited a lot of design decisions based on the 3300 interpreter.

Useful links: http://www.wang3300.org http://www.wang3300.org/emu.html (windows emulator for the wang 3300)

Disclosure: I created the wang1200.org and wang2200.org websites and the wang 2200 emulator already referenced by the wiki entry. I also created the wang 3300 website and emulator just mentioned. Goiter (talk) 21:12, 1 August 2012 (UTC)

Wang 2400 -- missing machine

The Wang 2200 was the second computer I ever programmed. The Wang 2400 was the third. It was very similar to the 2200 but with a floppy and double the RAM. Newark High School in Newark, Delaware, USA owned three or four 2200s and one 2400 in the 1970s. They were built into small tables and had cassette tape drives that took standard audio tapes. I still have a couple tapes somewhere.

Old, unsectioned comments

I have a WANG OIS with boards dated well pre 1980 and it has the previously labelled "Circa 1990" logo on it, so I have changed the date.


Wang was surprisingly unconcerned about the exact appearance of their logo and there were minor variations.

I believe that in later years the logo was more frequently rendered as a solid blue oval with white lettering on it; that is, the word Wang was white on a blue background. In earlier years the word Wang was brown on a white background with a brown oval outline surrounding it. That's what I think, anyway. If your OIS boards are the blue version then I stand corrected. Dpbsmith 17:21, 10 Mar 2004 (UTC) Having served 7 yrs in US Marketing and 3 yrs in US Advertising in Lowell, I can say that the cartouche (it was never an oval) was controlled by a strict, documented policy as you would expect in any multi-billion dollar company. Of course, some departments, especially fields offices, did not always follow the rules of color, size, background, placement etc. Many of the logos can be found on the Wang alumni FB Page in memorabilia that has been posted. — Preceding unsigned comment added by Ethelgarland (talkcontribs) 17:51, 21 August 2013 (UTC)

In the Wang VS world the logo was (and is) most frequently seen A) on equipment and B) in manuals and other printed literature. The logo on equipment was usually molded in 3D, sometimes with the letters silvered, and the logo in manuals was usually in B&W -- a black oval with white letters. The only color rendition I actually recall wasn't original -- it was the one offered in bumper-sticker form by a logomeister on eBay. Come to think of it I have some of the company coffee cups. One has blue with white letters, another has white with black letters. --Tjunker 09:35, 1 August 2005 (UTC)
As a former employee, 1973-1989, I thought I could clarify the logo situation. The logo was changed to the oval at the end of the sixties and was to be white letters on a blue background. There were various implementations of this unitl a corporate branding campaign in 1980, launched in conjuction with an outside ad agency, specified the font, color and demension of the oval. --Dr Glossary 15:50, 27 February 2006 (UTC)

Wang is considered a minicomputer vendor. None of their machines was known as a mainframe. I'm considering changing the section header to reflect that. Dyl 17:35, Aug 8, 2004 (UTC)


I agree with you. When I wrote the article I gave that section the heading "IBM-compatible minicomputers." On 10 March, User:Tjunker changed it to "IBM-like mainframes." I'm not sure why he did that, but I didn't think it was worth fussing over. Why don't you go ahead, but do put "See Talk" in the edit summary. Tjunker doesn't have a user page; if he did I'd consider asking him about it there, but since he doesn't I'll assume he either a) has this article on his watchlist and can join this discussion here if he cares, b) doesn't care, or c) is inactive...
Of course, personally I've never understood this use of the word "mainframe." To me it always meant the bay or bays containing the CPU in any floorstanding computer. One day I referred to "the mainframe" in this way and somebody said "O no, it doesn't mean that, it means a computer that costs over a million dollars" or something like that. I THINK the evolution probably took place in the IBM world. I'm guessing that third-party vendors of plug-compatible peripherals for IBM computers didn't want to say "IBM," so probably used "the mainframe vendor" as a circumlocution. Or possibly to mean "IBM and Amdahl." So then people started thinking that "mainframe vendor" meant "manufacturer of big honkin' MIS computers" or "IBM and the seven dwarfs," and then "mainframe" meant "big honkin' MIS computer." [[User:Dpbsmith|Dpbsmith (talk)]] 12:26, 9 Aug 2004 (UTC)
Yes, in my understanding of the word mainframe, it has multiple meanings:
  • A computer system that costs at least $1M dollars
  • A computer system that runs one function for a large company (say more then 200 people)
  • A computer with the highest level of reliability, availability and servicibility features
  • An IBM 360/370/etc compatible
Wang computers were none of these. A more typical application for a Wang machine was more likely word-processing for a small department. Anyways, the scale of the jobs & number of users for a minicomputer and a mainframe are easy to distinguish. Also, if you placed an IBM AS/400 physically next to an IBM System/370, it would be immediately obvious. Dyl 16:06, Aug 9, 2004 (UTC)
After doing a google search, I see the people who are selling these machines , are the ones referring to them as mainframes. People who are/were users/owners call them minicomputers. I'm going to make the change now. Dyl 19:06, Aug 10, 2004 (UTC)

As someone who has worked with Wang VS systems for more than 20 years and as the keeper of The Unofficial Wang VS Information Center I vehemently disagree. By any reasonable standard the VS is a mainframe computing system. The term may originally have had something to do with the physical structure of large computers and prices over $1 million, but over time the term has come to mean simply the type of large, multiuser data processing system typified by the IBM 360/370/390.

I was around when minicomputers evolved, and they were distinctly different than hefty data processing systems, usually characterized by shorter word lengths, lighter weight I/O subsystems, and operating systems more suited to message switching and laboratory equipment interfacing than to business data processing. The first minicomputers had no operating systems at all and were custom programmed from cold metal to do custom jobs. The first machine referred to as a minicomputer was the 1960-ish DEC PDP-1, which at a mere $100,000 was a price breakthrough. The fact that it had no operating system didn't get much attention, but was the reason the machine never did much in the way of traditional business data processing. By 1970 there were about 300 companies making minicomputers, most of them 16-bit and most of them with utterly no business DP software.

The 1977 Wang VS originally ran a slight superset of the IBM 360 instruction set, implementing linked list instructions and a stack architecture. It used the 360's base/displacment addressing system. When IBM released the 370, the VS microcode was updated to take advantage of some of the new instructions. The VS never ran any IBM software because Wang had a better idea: instead of trying to rivet an interactive layer on top of an IBM-style batch operating system, as IBM did, Wang built an interactive OS from the ground up, but also equipped it with features to facilitate serious batch operations: a Job Queue fed by background job submission and utilizing predefined tasks calle Background Task Initiators, and a Print Queue fed by spooled print output. Both supported Classes, such that submissions didn't have to specify anything more than Job Class or Print Class. Predefined task initiators and printers supported "scheduling" parameters that included the Classes they accepted. For instance, assign Print Class "I" to printers loaded with the Invoice form, then spool invoice output to Class I. The Print Task will dispatch it to any available printer that accepts Class I. Submit a background job to Class X and one of the set of Background Task Initiators that accepts Class X will get it when it comes to the head of the queue of Xs. That is pretty much the way IBM mainframe OSs have done it, and is distinctly different than the ways that minicomputers have handled workloads and printing.

In principle, an interactive OS does batch operations as a subset of interactive operations, but a batch OS can't easily be made to support interactive operations. Wang's design was successful. The VS excelled at both interactive and batch operations.

To limit the definition of "mainframe" to things that run IBM operating systems is erroneous. Since IBM relented after RCA built the first 360 clone and unbundled its OSs from its hardware, there has been a class of systems designed to run IBM OSs and applications. That class has been a subset of what could properly be called "mainframes."

  • A computer system that costs at least $1M dollars
This is a very dated definition. Even IBM "mainframes" many times more powerful than the 360s and 370s can be purchased today for much less than $1 million. The most expensive Wang VS ever fielded was the VS10000, with a list price of $1 million. More recent, more powerful models have ranged from $300,000 to $600,000.
  • A computer system that runs one function for a large company (say more then 200 people)
Mainframes have never been limited to one function. On the contrary, they have always been characterized by being multi-function, whether in the days of mostly batch processing or in the later days of interactive processing. The Wang VS is no different. It usually handles a variety of functions and has often been the "enterprise processor" in companies ranging to $500 million and beyond, and up to thousands of employees. I have personally worked with Wang VS systems supporting 700 concurrent users, and I know of systems ranging up to 2,200 users. This is not "minicomputer" country.
Nor does the extension of a model range down to very small systems disqualify the use of "mainframe." IBM has made very small S/390 models and has even packaged the S/3x0 chipset on an I/O board for use in RS/6000 and other systems as coprocessors. The range of VS models is characterized by complete software compatibility; there is nothing one can run on a large system that can't also run on a small system, assuming the same classes of devices are present if the program depends on special devices. All VS models are binary compatible.
  • A computer with the highest level of reliability, availability and servicibility features
Wang VS systems survive because they have long had unprecedented reliability, availability and serviceability despite not having taken on many of the more modern technological features now associated with those characteristics. There are many VS systems in production operation today that were manufactured in the 1985-1990 period when Wang was riding high. The 1985 VS300 is in service today and can be field upgraded to the fastest 1999 VS18950 CPU. VS failures have historically been rare. The machines were built to last and they have indeed lasted.
  • An IBM 360/370/etc compatible
That's not a requirement for a system to be referred to as a "mainframe." It was just a common case, due to the popularity of 360/370 clones, starting with the RCA 360 clone that started that trend. In fact, though, any IBM BAL programmer instantly recognizes VS assembly language, since not only does the VS have all the IBM instructions, but its macro assembler and linker are also patterned after the IBM mainframe tools.
  • Wang computers were none of these.
Wang VS computers were in fact all or most of these.
  • A more typical application for a Wang machine was more likely word-processing for a small department.
Not hardly. You must be thinking of other Wang systems that preceded the VS in the 1970s, such as OIS or WPS, or the BASIC-only 2200. Wang VS systems did WP as a shoe-horned afterthought. The VS was designed first and foremost to run data processing, interactively and in batch. It supports machine instructions that make COBOL virtually an assembler -- most COBOL verbs compile into single VS instructions. It supports machine data types of halfword and fullword integer, character, binary floating point, decimal floating point and packed decimal. Yes, it does arithmetic with decimal numbers, which is another significant characteristic of mainframes. Yes, it does decimal floating point, which can be used for business calculations and produce results free of binary rounding errors. Some major business systems were written in VS compiled BASIC using decimal floating point for dollar amounts.
Some examples of how the Wang VS has been used in my experience:
  • Enterprise system for a $600 million distribution company, 700 users
  • Enterprise system for a $250 million manufacturing company, 400 users
  • Enterprise system for a $100 million insurance company, 50 users
  • Enterprise system for a New Zealand bank
  • Enterprise system for a major mortgage company, 400 users
  • Major banking operations at a large NY bank
  • Major banking operations at a large S. Korean bank
  • Wang VS systems presently run the Colorado Lottery
  • Anyways, the scale of the jobs & number of users for a minicomputer and a mainframe are easy to distinguish. Also, if you placed an IBM AS/400 physically next to an IBM System/370, it would be immediately obvious.
S/370 has long been obsolete. There are probably modern S/390 systems smaller than some AS/400 models. Physical size is no longer a reliable indicator. But if you want to talk physical size, the VS10000 and VS12550 were about 5' tall, about 5' wide, and weighed in at about 900 lbs, with backpanel connectivity for hundreds of coax workstations and printers and TC devices and dozens of disk and tape drives. Wang VS SCSI comes out of large VS models as High Voltage Differential, to permit locating disk and tape enclosures up to 82.5 feet from the VS system unit, because in large computer rooms the storage often was located rows away from the system units.
While it's true that IBM mainframes extend higher into the heights of users and horsepower, the VS has commonly served several hundred users, with complements of I/O such as 20-60 disk drives, 20-100 printers, 10-20 tape drives, and dozens of synchronous communication links to other VSs and to IBM mainframes, all in a coherent environment that facilitates connecting large numbers of users to applications. Support for multiple file types is integrated into the OS. Print and text files are compressed by default. 17-way indexed files are standard. Rollback and rollforward recovery and clustering of multiple systems are integrated into the OS. The VS does SNA and LU 6.2, and virtual 3270 for screen scraping IBM session links. I've worked at several VS sites where printing volumes exceeded 900,000 pages per month on laser printers, including the large Xerox ones.

In contrast, minicomputers started with the 18-bit PDP-1 and progressed through mostly 8-bit, 12-bit and 16-bit, eventually reaching 32-bit, but usually with none of the hardware or software infrastructure that would allow a business application to be built and operated without considerable additional programming beyond the application itself.

BTW, I don't have a Wiki user page because all my time is consumed in activities related to the Wang VS. I admire those of you who make substantial contributions to Wikipedia but that isn't my thing.

Also BTW, I added the newsworthy paragraph about the rebirth of the VS but had some difficulty not making it sound like an advertisement. Comments on how best to do this are welcome.

--Tjunker 09:35, 1 August 2005 (UTC)

OK, I hope you don't mind... the New VS is a very relevant and worthwhile addition to the article, but I did shorten it and tone it down. I see that tjunker@transvirtual.com system is the listed contact for transvirtual on the press release. In point of fact, I think anyone interested in migrating a VS system will understand the significance of the product without needing it to be spelled out.
The cited press release does not mention the name "New VS" or, in fact, any name for the product.
I replaced the tinyurl with a full one. I am not quite sure whether or not there is any policy on this, but in the past I have seen Wikipedians take exception to the use of tinyurl links. The issue is that you have no way of know where the link will take you until you click on it, which affords a possibility of pranks and vandalism.
How, exactly, does one read a "standard VS backup tape" on a PC?
I'd be happier if there was some sort of source citation or other backup for the statements to the effect that "Wang VS conversions (or "migrations") have historically been fraught with difficulties." No doubt you're referring to migrations that are based simply on porting COBOL.
I assume that the New VS does not embody any 928 hardware or any way of connecting directly to 928-based terminals. How is migration of terminal-based systems handled? A standard LAN with some kind of Wang workstation emulator running on the remote terminal, or what? Dpbsmith (talk) 11:04, 1 August 2005 (UTC)


Fair enough. I don't mind.

On the subtopic of spelling things out, you might be surprised at just how things have to be spelled out in detail for VS customers to grok what the New VS is, but I agree that wiki is not the place to do that.

The press release was drafted before the name was settled on. I realized that when I wrote the wiki text but left it as I wrote it since the product is now known as the New VS and the data sheet will soon be up on the TransVirtual site. The wiki text could use a bit of massage to avoid suggesting that the New VS name is in the press release by explaining that the name came about afterward.

TinyURLS: good point. I created them to use in emails because some people have trouble with mail clients, how they present URLs, and especially in copying fragmented URLs to browser URL boxes.

Reading a "standard VS backup tape" on a PC is easy. The reading, of course, is done by VS software, so the contents of the tapes are thoroughly understood. All VS tape hardware is either already SCSI or there are alternative SCSI drives available to handle the same types of tapes. Because the Linux 'st' ("SCSI tape") module is so badly broken, most projects with a need for serious SCSI tape operations have chosen to write their own tape handling. TransVirtual has done that, too, and that means that any wrinkles found in drive variants going forward are easily accommodated. 4mm was the first tape type supported, and appears to work for all generations of 4mm. Next came DLT support, and it, too, seems to work for all generations of DLT. In both cases the New VS supports drive generations far advanced beyond what the VS supports in addition to handling VS drives and tapes and compatibles. 9-track, which traditionally was in the form of tri-density Telex and Kennedy upright drives with Pertec interfaces, is easily replaced by tri-density front-autoloaders of the HP 88780 type, which are widely available and fully maintainable. Streamer tape drives like the 150 MB Viper found in the VS5000/6000 boxes can move right over to any Unix/Linux system and are plentiful in VS circles.

I should mention, though, that the New VS is not available on just any PC. The product would suffer from bad customer choices and the support issues would be nightmares. It is available only on selected Linux/Unix servers with very specific configurations. The primary platform is the Dell PowerEdge 2850 with a base configuration of dual 3.6 GHz Xeons w/1 MB cache, 1 or 2 GB main memory, DRAC (Dell Remote Access Card), PERC/4e RAID, and a minimum of a pair of mirrored U320 drives. Since VS memory can range up to 2 GB, very large VS replacements would call for more main memory on the PowerEdge. Most VS systems have in the range of 10-30 GB of VS disk volumes, so the PE 2850's present upper limit of 1.5 TB of usable RAID5 storage far exceeds most requirements while staying entirely within the PE 2850's 2U enclosure.

The New VS is also running on the IBM OpenPower and will probably be offered on that platform even though performance is not what had been hoped and the upper performance tiers probably won't be offered. Next to be examined is IBM xSeries (Intel), which in theory should perform like the Dell Poweredge.

The New VS now runs on Dell PowerEdge and IBM xSeries. The IBM OpenPower wasn't up to snuff and we disqualified it. TJunker 13:37, 13 October 2006 (UTC)

Citations on Wang conversion difficulties: there is little in the public record, but much in the experience of those of us who have worked in the VS field for many years.

The superlative crash-and-burn VS migration was that of Kent Electronics, a former client of mine although I had nothing to do with their migration. They hired Sun Microsystems as the prime contractor and DevTech, I think, which later became NeoMedia, was a subcontractor. I don't recall the COBOL vendor but at that time their indexed file system came from a third party, so there were four significant players in the project. After all the conversion work was done and the project went into live load testing as the last phase before cutover, the Sun Solaris system bogged as users logged in from branch offices around the country. It went fine until a certain user threshold, something like 175 or 275. Thereafter it got so slow that minutes elapsed between screen reponses. All the players tried to figure out the problem and the test was repeated several times with investigations inbetween that failed to fully identify the cause of the slowdown. I heard that Sun eventually sent in OS experts, but that they, too, failed to figure it out. Sun was eventually invited to leave, and on their way out to purchase another large VS for Kent or see them in court. Sun did, and Kent went back to growing their VS cluster to be three large VSs... a VS16850 and two VS12550s, supporting about 700 concurrent users. The Kent IS manager lost his job and a CIO was hired, and after a few years they regrouped and did an India-based rewrite to something like Informix on AS/400 at a huge cost in money and with complete loss of control of the source code by in-house staff. The company was then bought by Avnet and the entire new system was tossed in the trash. The retired VS equipment, once on the books at over $1 million and still around, stuffed away in unused rooms, was sold to a scrapper for 30 cents per pound.

Some migrations or replacements, such as the one done by Mellon Mortgage Company, were rated as successful by the suits but not by the people actually working with the systems. In other cases everyone, including the suits, knows that the new system sucks, but it's impolitic to admit it, so everyone grits their teeth and claims it's great. A walk through the user cubicles often reveals the truth in both those types of cases.

All the long-time VS consultants I know have personal knowledge of less-than-satifactory VS migrations. I have heard of a few supposedly wildly successful migrations, but only second or third hand.

928 hardware: (for those who may not know, "928" is the dual-coax and twisted pair technology used to connect Wang VS workstations, printers, and various other outboard peripherals, including telecommunication interfaces, low-density 9-track tape drives, streamer tape drives, UPS interfaces and a host of other miscelleanous things. It uses a differential signal at about 4.5 Mhz and multiple layers of protocols, depending on the type of device. The I/O coprocessor that runs 928 ports is microcode loadable, not only for its own multithreaded, embedded OS but for port-specific microcode modules tailored for specific device support. The minimum number of ports supported on a single I/O coprocessor is 32; the maximum is 128.)

The original plan for the New VS included developing a 928 PCI card to use the already-existing VS5000/6000 24-port expansion box, which has an extremely dense D-sub snap-on connector on the VS side suitable for a PCI card L-bracket and offers 24 standard VS dual-coax 928 ports in a free-floating box at the other end of the cable. The original reason for the plan was the clear necessity of supporting Lightspeed NVS gateway PCs, which use Wang Local Office Connection (WLOC) cards, which connect to VS dual-coax 928 ports. A side benefit was to be support for the odd VS workstation or VS printer found at most VS sites.

Late in 2004, though, TransVirtual concluded the acquisition of the Lightspeed NVS product line, making it possible to integrate the gateway PC functionality directly into the NEW VS and support LS NVS desktops directly via TCP/IP without gateways and without 928 for WLOC cards. So the urgency to develop the 928 PCI card diminished. It's still in the plan, but has slipped way down on the priority list. There are other things that are much more important, such as async to support existing async printers, TC to support WSN and SNA, and RSF, to cluster New VS instances over TCP/IP whether in the same or different boxes. Those projects are now underway and 928 is on the back burner.

928 is now supported on the New VS. TransVirtual now manufactures a PCI Universal Serial IOC (USIOC) that presents the same 37-pin MUXBUS connector as the legacy USIOC. Externally, traditional VS connectivity equipment is used to mount coax, twisted-pair and fiber optic Active Port Assembles (APAs) and 6550/6550A telecommunications APAs. This arrangement allows New VS customers to retain needed or beloved 928 devices, and such sites are the rule rather than the exception. Imagine a site with a Wang 2000-lpm band printer, a warehouse of five part forms, and an organization solidly keyed to handling the differently coloered copies. Rather than tell them how good laser printing is and that they have to change their workflow, we can now plug the band printer into the New VS and minimize the impact (pun?) on the customer's organization. TJunker 13:37, 13 October 2006 (UTC)

Terminal-based systems: During the 1990s almost all VS sites moved from native Wang VS workstations to Lightspeed NVS, which runs workstation emulation on standard desktop PCs and connects to the VS through a PC gateway. The gateway in turn connects to the VS via 928 or SCSI. Since VS sites already use Lightspeed, the issue was not so much how to replace Wang workstations but how to support Lightspeed desktops. LS NVS does logon service, file transfer, printing, Web virtual terminal, and has APIs that allow the VS to control actions on a PC and a PC to make things happen on the VS. By about the middle of 2005, Lightspeed connectivity was supported in the New VS. Since the LS NVS gateway is really just a mux/demux and has no particular awareness of the traffic it handles, replacing it with a module in the New VS supported all the functionality, which is resident in the VS and in the client PCs.

So the answer to the question of how users connect to the New VS is that they connect the same way they connected to the legacy VS, using the same desktop PCs and the same Lightspeed NVS software.

Seamlessness: I don't use the word lightly. A legacy VS SCSI disk can be attached to the New VS and the VS OS IPLd and the system brought up. Or VS disks can be imaged into files in the New VS. Or entire systems can be restored from VS backup tapes. Lightspeed desktops can connect just by changing the gateway IP address in their configuration. Archives of backup and other data tapes can remain accessible just by having a compatible drive on the New VS, either brought forward from the legacy VS or bought new to gain greater life and perhaps better self-cleaning technology. The Support Control Unit (SCU) admin interface is fully supported except for hardware diagnostics that are no longer relevant. New SCU screens provide for mapping of what the VS sees as "physical devices" to host platform devices or files. New Control Mode debugging features allow multiple breakpoints to be set for execution addresses, opcodes and other runtime attributes. Note that Control Mode debugging is for OS programming and diagnosis, not for normal programming. Low-level device operations such as disk initialization and formatting operate seamlessly. Volume Table of Contents (VTOC) analysis and validation is seamless. Disk, tape and workstation hardware tests work seamlessly. The VS OS and all its tools really think they are still running in a legacy VS.

More on tapes: Many VS sites have used DAT, DDS or DDS-2. A few have used DDS-3 and Wang may have released a DDS-4 but if it's in the field it's very rare. Putting a good DDS-3 or DDS-4 drive on a New VS can provide backward compatibility with archives of DDS-2, DDS and even DAT tapes while likely working better and delivering a longer life than an older drive and technology. New backups on suitable media would be much faster than before.

Wang released a DLT drive, using the Quantum DLT 8000, usually found only on large VS systems. Since VS SCSI was slower than the DLT 8000, they recommended using DLT III tapes to prevent shoe-shining. Unfortunately, the DLT 8000 is long out of production, and later drives can't read DLT III tapes. Some can read DLT IV tapes, though, which the DLT8000 drive can also write. And many serviceable DLT 8000 drives are available in the resale markets and some may even still be in slow distribution channels.

There is a New VS undergoing evaluation right now (Aug 2005) at a VS18950 site. Files are transferred on DLT IV tape, written on a DLT 8000 drive on the VS and read on an SDLT 320 drive on the New VS. Should a VS BACKUP be done on the New VS to the SDLT 320 drive using Super DLTtape I media, it runs at SDLT 320 speed, not at old VS DLT 8000 speed. So in some cases a single piece of modern hardware can both read the legacy media and perform on its own at many times the legacy speed and capacity.

New VS systems are now being shipped with DDS-4, DAT72, SCSI 9-track and SDLT320 tape drives, all of which work perfectly with various VS backup/restore tools. TJunker 13:37, 13 October 2006 (UTC)

More on the 'mainframe' vs. 'minicomputer' issue: All VS models from small to large have a standard option to use up to 28 CATV channels as 10-Mbps LANs on an up-to-6,000-foot campus broadband backbone system called Wangband. All VS models have standard options to support multiple 32-peripheral clusters, over direct fiber or via the broadband backbone. All VS models were designed to have their workstations or terminals located up to 2,000 feet from the system unit long before ethernet and LANs came on the scene, and the large form factor models to have their disk subsystems up to 82 feet away. All VS models support optical disk document imaging subsystems, combined DP/imaging workstations, and image annotation. The VS integrated development environment standardly supports Assembler, COBOL 74, COBOL 85, BASIC, RPG II, C, PL/I, FORTRAN, Glossary, MABASIC and Procedure (what would be called a scripting language in *nix systems). PASCAL is also supported for I/O coprocessor development. From the integrated Editor one edits, compiles (if a compiler language), links (if appropriate), and runs programs under development. The PACE 4GL and database had bulletproof referential integrity built into the Data Dictionary level many years before Oracle figured that out. VS clustering supports distributed databases with two-phase commit, file sharing across the cluster, and Print and Job class distribution across the cluster. VS/VM, the Virtual Machine OS, allows IPLing and re-IPLing multiple copies of the VS OS in a single machine, and allocation of devices to virtual machines down to the workstation window level (e.g. it is possible to have the four windows of a MultiWorkStation unit allocated to four different virtual machines). VS/VM supports virtual clustering.

Have I made my case for 'mainframe?'

--Tjunker 06:49, 6 August 2005 (UTC)

  • No, you've certainly given cogent reasons but I'm not personally convinced. I find your comments in general above very interesting. I've tried to boil some of it down to a couple of sentences to add to the description. Dpbsmith (talk) 14:35, 6 August 2005 (UTC)

Cogent reasons, but you're not personally convinced. I don't follow that at all. Mellon Bank had a network of 45 VS systems running Wang Office serving 16,000 users and you think that's a minicomputer? I really don't know where you get your definitions nor where you have been. Minicomputers, historically, have never been business data processing systems except with a lot of specialized development and pain. VS systems have primarily been business data processing systems in the mold of IBM 360/370, out of the box, with IDE supporting about a dozen languages. WP never figured very heavily in the VS market. A key is hardware doing decimal arithmetic, which "minicomputers" have almost never done. Maybe you'd like to try camping out for a week in a VS10000 or VS12550 cabinet and then tell us it's a minicomputer.

There's an interesting paper by an outsider that tries to explain Wangnet, the broadband campus networking system fielded by Wang in the early 1980s and in use to my knowledge at least until 2000. It's another much-more-than-minicompter aspect. Using CATV technology and supporting up to 2,200 intelligent devices as well as large numbers of multiplexed communication circuits from a backbone that could run up to 6,000 feet, it was decidedly more than what minicomputers fielded for interconnect subsystems. Both Exxon Company USA and Mellon Mortgage Company in my personal experience had Wangnet backbones running up through their Houston buildings and branching to encircle each floor as the interconnect for hundreds of users and scores of printers throughout their buildings.

From an early 1990s Wang VS brochure created, I think, in Ireland: "The VS is so close architecturally to an IBM mainframe that one of the world's leading database suppliers uses a common set of object code for both architectures."

BTW, someone introduced this into the article, which is decidedly false:

Though built from commodity hardware, the requirements for very specific configurations require that the hardware platform be provided by Getronics.

The specificity of configuration comes from TransVirtual and is simply a minimum base of a Dell PowerEdge 2850 that includes dual 3.6 GHz Xeon processors, PERC4/e RAID, DRAC, no less than a mirrored pair of drives, and at least 2 GB of main memory, to ensure that we can remotely support the systems, that they will be reasonably fault tolerant, and that they will have the capacity to run the full range of VS equivalents and be remotely upgradable as to performance. There is no requirement whatsoever that Getronics provide the platform. In the U.S. TransVirtual sells alongside Getronics and either provides the platform or allows the customer to provide it per the minimum specs or, if it's a Getronics sale they may provide it, obtain it from TransVirtual or allow the customer to provide it. Outside the U.S. either the local Getronics subsidiary or affiliate or the customer or the customer's facility management vendor can provide the platform.

--Tjunker 12:50, 20 August 2005 (UTC)

In section ‘Word Processors’, does “Z80 and 65K of RAM” actually mean 65536 bytes of RAM, i.e. 64 kilobytes? (The Z80 uses 16-bit addresses, so can only address 65536 bytes of RAM without resorting to tricks like switching banks of RAM or using I/O to access external RAM as a device.) If it does use such a trick, then I'd consider that as interesting enough a detail to mention explicitly.

--pjrm (of no wiki account)

Yes, full-featured Z80-based Wang DP/WP workstations maxed out at 64 KB -- 65536 byets of RAM. Wang used the Z80 extensively as a microcode-loadable subsystem, the original idea having been to offload work from the host system. Z80s were used not only in workstations but in printers and other peripherals and in internal subsystems inside the VS.

DP and WP on the VS are entirely different, with microcode for one or another being loaded into the workstation on demand. DP mode is roughly like IBM 3270, a block-mode workstation with field attributes and modifiable and non-modifiable fields, with nothing going back to the host until Enter or one of the 32 function keys is pressed. WP mode turns control over to the workstation, with the host becoming little more than a document server. The keys are remapped and instead of DP function keys they become things like Indent, Center, Search, Replace, Copy, etc. The Enter key becomes a newline key for the document text being edited and the Execute (End) key is used instead to invoke things.

The WP microcode works with segments of a WP document, shuttling the segments to and from the host as needed. In a global search and replace, for instance, the workstation microcode successively requests the document blocks from the host, makes the requested changes, and returns each block to the host. Some WP functions are actually performed by host programs run by the workstation microcode, but all the strictly WP document editing functions are performed by the workstation microcode in no more than 64 KB on documents essentially unlimited in size.

-- TJunker 11:43, 11 February 2006 (UTC)

I hope I have clarified the minicomputer/mainframe distinction in VS product lines in this recent edit. I won't repeat all that I've said before, but I'll opine that anyone who believes that the VS was/is not a mainframe probably has had experience with Wang systems limited to OIS and/or 2200, which could more accurately be described as minicomputers.

For anyone who may be interested, sales of the New VS -- the VS22000 -- have begun in earnest, replacing legacy VS systems in the U.S. and Europe. Since the last flurry of discussion here there have also been been further technical developments.

WSN (Wang System Networking) is now working and being released to the field. It works over TCP/IP instead of legacy synchronous lines and appears to be stable over the Internet.

RSF (Resource Sharing Facility), the VS clustering feature, was just completed and is being released. It, too, uses TCP/IP instead of legacy FDDI.

Virtual PIB, which replaces the hardware coax-based Wang Printer Interface Box, is working in the field, supporting both parallel printers and network printers. At least some Wang coax (928) printers can be field converted to parallel port operation to allow them to be moved to VPIB on the New VS. VPIB is fully microcode-loadable and supports all printer variants supported by the original PIB. Target printers don't have to be legacy Wang printers, though, and Linux print queue and CUPS in the VPIB are alternatives to running direct-connect parallel printers.

Direct support of Lightspeed NVS desktop clients was integrated in early 2005 in the wake of our acqusition of the LS NVS product line in late 2004. LS workstation emulation and file transfer are much faster now with the elimination of the PC gateway formerly required for connection to the legacy VS.

928 coax support is in engineering and should make its appearance in 1Q or 2Q of 2006. Supporting 928 is a shortcut that will bring immediate functionality of SNA, X.25 and other peripherals and protocols by using legacy outboard TC units on the New VS. Connection of the New VS to VS5000/6000 IOCs will follow 928, providing seamless Wang 802.3, Image Transfer and other neat things. This stuff is necessary because almost all larger VS sites have one or another requirement for things like SNA or 802.3, and it's faster to interface the legacy outboard units than to virtualize all of those controllers and devices with all of their microcode variants. Eventually there will be software replacements for the interim legacy peripheral units.

928 support became available early in 2006 as expected. It is now in the field in multiple New VS systems, supporting Wang workstations, printers and telecommunications. TJunker 13:37, 13 October 2006 (UTC)
A new development is a multichannel software pipe between the VS and Linux environments. A customer doing alpha testing of the new facility has successfully queried an Oracle database from a VS COBOL program and received the correct result data. The pipe is not specific to communicating with Oracle, though — it supports multiple data and control channels between modules in the VS environment and modules in the Linux environment. Such a pair of modules can support any kind of data transfer or interaction for which it is written. TJunker 13:37, 13 October 2006 (UTC)

A live test earlier this year in Germany (mentioned earlier as being in progress) switched over 700 users from a production VS18950 to a New VS. No problems were encountered and the original two-day test was extended to a full week. After the test the production files were moved back to the legacy system. During the test a technician caused the site UPS to drop power twice. Equipment failed to come back up on another production legacy VS and on a network server, but the New VS came back up both times with no loss of data other than incomplete transactions still up on screens at the moment of power failure. No virtual VS disk volumes or files were damaged.

The live test in Germany was followed by a substantial purchase. There are now multiple New VS customers and systems in Germany. The largest runs 550 users on a New VS. TJunker 12:54, 13 October 2006 (UTC)

Performance on the standard dual 3.6 GHz Xeon Dell PowerEdge presently tops out in the range of 8800-9300 FAST, where FAST is a Wang benchmarking system that applies only to VS systems. It's being pushed to 12000 in coming months. By comparison, the original VS80 from 1978 is rated at FAST 100, while the fastest legacy system, the 1999 VS18950, is rated at FAST 6300. Disk throughput, though, is very much faster, since the legacy VS systems are stuck at SCSI-1 (5 MB/sec). We recently broke SAM, the System Activity Monitor tool, when we did more than 9,999 virtual VS disk I/Os per second running a VS disk exerciser utility on a New VS.

-- TJunker 12:11, 11 February 2006 (UTC)

SAM has since been fixed by Getronics and now can register over 9,999 disk I/Os per second. TJunker 13:37, 13 October 2006 (UTC)