Talk:History of IBM CKD Controllers

2310

The 2310 was not part of the 2314 family; it was a cartridge drive. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:04, 5 January 2016 (UTC)

That is an obvous typo, why don't u just fix it instead of raising it as a talk issue? I leave it to you to fix it 17:44, 5 January 2016 (UTC)

Merge 2841 article into section of this article

Why not, they cover pretty much the same stuff? Tom94022 (talk) 07:33, 7 January 2016 (UTC)

I agree, but there is some additional historical information in the IBM 2841 article that should be added to this section before deleting the article. John Sauter (talk) 03:58, 13 January 2016 (UTC)

Done Tom94022 (talk) 04:40, 13 January 2016 (UTC)

You didn't merge the contents of the 2841 article into the history article before deleting it. Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:49, 14 January 2016 (UTC)

Actually I believe I did so, leaving out material that was redundant to the rest of this article, but if you fell I left something out please go ahead and add it here rather than reversing the merger. Of if u prefer talk about what needs to be changed in this article. Tom94022 (talk) 22:32, 14 January 2016 (UTC)

Split 2841 material into IBM 2841 and restore original content

The original approval for merging the IBM 2841 article into History of IBM CKD Controllers was contingent on preserving the contents of both articles in the merged text. This was not done. Accordingly, I propose splitting the articles and reinstating the original contents of IBM 2841. Shmuel (Seymour J.) Metz Username:Chatul (talk) 21:32, 7 February 2016 (UTC)

Why don't u just add the material u think is missing? Note most of the material left out was IMO redundant, but there seems to be little reason to have two 2841 articles so why not just fix the one article? Tom94022 (talk) 01:34, 9 February 2016 (UTC)

That's a disingenuous question, given you propensity to capriciously revert my changes. I'm tired of edit wars and would rather raise the issues and let someone else make the changes. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:17, 10 February 2016 (UTC)

BTW, the capacities of the attached devices were left out of this section for consistency with the rest of the sections in this artilce given the information is available in the links to the devices. I suppose we could add min-max capacity to all the controllers with assorted devices in all the sections but IMO that is IMO unnecessary detail

Furthermore, the suggestion to split does not meet the content splitting quidelines. It is one subject which should be treated in one place. Tom94022 (talk) 01:53, 9 February 2016 (UTC)

It's not just the capacities that are missing, it's the list of original devices and the references. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:17, 10 February 2016 (UTC)

I left out the references as redundant either to this article or to the HDD article.. Again there is no justification for splitting, if u can improve the section in this article please do so; in the meantime please stop leaving cryptic criticisms. Tom94022 (talk) 17:35, 10 February 2016 (UTC)

The reference manuals for the IBM 2841 are hardly redundant. The fact that your list is incomplete is hardly cryptic; you had only to look at the original article to see that it included the 7320, and you acknowledged that you were aware of the missing references. I will happily do the edit, but only if you will commit to not reverting it. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:39, 10 February 2016 (UTC)

The 2841 ref manual is already in the article as ref 3; if u want to go ahead and link in the 2841 section to the existing ref please do so. I'm not sure we need more than one 2841 reference manual but if you do we shall see what it adds to this section. The 7320 is there as a note to the 2303 which IMO is all that it deserves since very few shipped and it was replaced by the 2303. I rarely revert but I do correct and or simplify so I really don't know what I will do about any edits until I see them. Tom94022 (talk) 22:43, 10 February 2016 (UTC)

While correcting a typo I added a redundant link to the 2841 ref manual in the 2841 section (it was already linked in Section 1). I remain curious as to any real improvements Chatul might want to make to the article. Tom94022 (talk) 17:53, 13 February 2016 (UTC)

Your text lists only one edition of the 2841 manual. Two editions are relevant because the 7320 appears in the original edition but the 2303 does not, while the 7320 does not appear in the new edition. Your text is also missing the 2841 CE manual.

The 7320 was part of the original 2841 announcement and was in the original 2841 manual; the 2303 was neither. As such, it is ahistorical to list the 2303 but not the 7320. Shmuel (Seymour J.) Metz Username:Chatul (talk) 21:18, 18 February 2016 (UTC)

The 7320 was not listed in the -2 rev of the 2841 manual dated 10/22/1965! Furthermore, Datamation is a reliable source that the 7320 was replaced by the 2303 on or before March 1966. The first high end S/360 shipped in November 1965 so it likely very few if any 7320s ever shipped on System/360. Accordingly it is apprropriately treated as a footnote to the 2303 and linkage to the the earlier ref manual is unnecessary. More detail would give undue weight to this device. Tom94022 (talk) 07:50, 19 February 2016 (UTC)

Of course it as not listed in the wrong edition; what matters is that6 is was listed in the first edition. If you are saying that only summary information for the controller belongs in History of IBM CKD Controllers, that is an argument for splitting it so that more detail can be present in IBM 2841.

As to the importance of the 7320, it was present in the 2841 announcement and it is ahistorical to pretend otherwise. Regardless of popularity (I suspect that the 2301 was more common), the 2303 was a Johnny-come-lately. Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:19, 19 February 2016 (UTC)

At the risk of repeating myself, more detail would give undue weight to 7320! Phister shows 670 head per track devices installed on IBM systems in 1964 decreasing to 650 in 1965 when the s/360 began shipping and then increasing rapidly thereafter to 2,480 by the end of 1968 (This is all model head per track, drum and disk). By October 1965 7320 was not listed in the 2841 manual which means it was probably no longer offered on the 2841 and by early 1966 Datamation reported it had been replaced by the 2303. How many 7320s do u think actually shipped on the 2841 given its very short availability? Again according to Phister there were about 800 s/360s shipped in 1965 (most of which did not have HPT devices) but over 4000 systems shipped in 1966 by which time the 2303 had replaced the 7320 I suspect given the 2303's much lower price (lower than either the 2301 or the 7320) that the 2303 was far more common Given the timing, price and looking at the actual installations it is likely there were few if any 7320s installed on the 2841. Rather than a "Johnny-come-lately" the 2303 was likely the dominant fast access device on the S/360 until the 2305. The 7320 was a footnote and that is how it should be treated in this and any article.

We are editors, not transcribers; footnoting the 7320 is a reasonable editorial decision. Is there any other detail u think missing, more detail on the already footnoted 7320 hardly justifies a separate article? Tom94022 (talk) 20:53, 19 February 2016 (UTC)

RfC: Should IBM 2841 be a separate article and should it list the IBM 7320

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

---

The issues in dispute are whether History of IBM CKD Controllers #2841 should be split back into a separate IBM 2841 article and whether the list of supported devices should include the IBM 7320 in addition to or in place of the IBM 2303, rather than relegating it to a footnote. IBM announced the 7320 first, but dropped it once they announced the 2303.

See Talk:History of IBM CKD Controllers #Split 2841 material into IBM 2841 and restore original content for previous discussion. Shmuel (Seymour J.) Metz Username:Chatul (talk) 22:24, 23 February 2016 (UTC)

Oppose

• There is no reason to distinguish this one of many IBM CKD controllers with its own article. The 2841 had an existing very short article and it makes sense to incorportate it as a section of this new article covering all such controllers much as is done in History_of_IBM_magnetic_disk_drives where many IBM disk drives are coverred in one article. The 7320 is coverred in the 2841 section of this article as a footnote an in-line note (using efn template) to the 2303 since all reliable sourses suggest it was a very short lived S/360 product with few if any shipments as a S/360 CKD controller. Any further treatment would give undue weight to this inconsequential device. Tom94022 (talk) 22:58, 24 February 2016 (UTC)
• The existing article seems to need work and splitting it would divide effort to craft a better article. Currently, this article does not seem like a history so much as just a collection of info. Seems like it could use work to avoid running a foul of WP:IINFO. Klaun (talk) 16:26, 2 March 2016 (UTC)
• Here by Legobot. I see no reason for a split. The section is relatively short as just a small collection of information that wouldn't qualify for WP:CONTENTFORK and doesn't appear to have standalone notability for a separate article. Kingofaces43 (talk) 06:16, 4 March 2016 (UTC)

In Favor

• Three of the 5 devices originally announced for the 2841 were reengineered versions of IBM 1400 and 7000 series devices, and that's an important part of its history. From that perspective the 7320 I more important than the 2303. Shmuel (Seymour J.) Metz Username:Chatul (talk) 23:10, 25 February 2016 (UTC)
• Include: This is a history article; one includes or excludes items because of relevance, not because one party finds it interesting and another finds it uninteresting. To assert that a device was inconsequential because it was a short-lived product argues a stunning lack of insight into the nature of technical or commercial computer or computer peripheral design or development, either for technical-historical or market-historical reasons. Factors influencing the introduction or retention of a device may include the availability or unavailability of components, the time scale available for introduction, customer pressures for delivery, staff availability when top design or engineering staff are preoccupied with rival projects, unexpected problems of compatibility or real-life performance, support or reliability and many other factors -- such items are historically important and have on occasion affected the course of IT history, decidedly including IBM history. Even when inside information on such variables happens not to be available at the time of writing, publishing what is available in context invites more contributions, and even before any such later revelations, presents one more item of substance for readers to develop their own ideas about what is consequential and to which factors they are of consequence. For editors to make such decisions for such readers is risky at best and largely unethical in most cases. As for footnotes, they are distractions and accordingly are to be avoided whenever practical; they more often are a sign of bad writing than of insignificant material. They are not to be reserved for inconsequentialities, but for decidedly relevant material that is out of context, but parenthetically too important to omit, such as editorial comment, explanation, or updated information in a narrative history. In the days of tree-butchery and manual typesetting, footnotes commonly represented stop-press information, but in electronic word-processing and storage media that doesn't apply, and in an encyclopaedia of this type, editorial comment is rarely permissible. JonRichfield (talk) 08:44, 2 March 2016 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Incorrect descriptions of 3830-2 and 3830-3

1. Neither the 3830-2 nor the 3830-3 had storage directors.
2. The 3830-3 could attach 3350 drives

Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:32, 22 March 2016 (UTC)

Here you go again, if you have reliable sources for your assertions, why not just make your changes in the article? Regarding the assertions above:

1. You are incorrect, both are storage directors that attach 3333s.
2. You may be correct, the source I referenced did not allow 3350s

Tom94022 (talk) 00:42, 23 March 2016 (UTC)

Here you go again. We do you kieep asking a question that I have already answered multiple times? My reason for not just making the changes is your history of capricious reverts, as you well know. Asking the question again will not change the answer.

1. Repeating the claim that I tagged as {{Cn}} is not a citation. On what do you base your incorrect claim that the 3830-2 and 3830-3 are directors.
2. As for the 3350, see, e.g., http://bitsavers.org/pdf/ibm/38xx/3850/GA32-0030-2_3850_Mass_Storage_System_Installation_Guide_May77.pdf

Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:04, 23 March 2016 (UTC)

I am hoping that you will stop the ad hominems and stop wasting my time. From now on I will one time request u just edit in the article and if you don't then the article remain as it is. Tom94022 (talk) 06:26, 24 March 2016 (UTC)

I am hoping that you will stop the hypocrisy and stop wasting my time asking questions that I have already answered. If you stop capriciously reverting my editis then I will be willing to Wikipedia:Be bold, otherwise I will continue to discuss changes before making them. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:40, 24 March 2016 (UTC)

You probably should have stopped with the {{Cn}} on the 3380 material; u were aware that your assertion was disputed and were asked for a reliable source. You were even given a rather cryptic reason why it was wrong. It then was rather presumptuous for you to make such drastic changes with such a thin justification (just because u can't find it doesn't mean something is incorrect); particulary so given your assertion was disputed.

The 3830-2/3 is properly categorized as director type and not as a conventional type storage control! The 3380-2/3 are not conventional storage controls in that they do not interface directly to a conventional drive (a B-Unit) as do the others so categorized. In that sense your categorizing them as "Conventional storage controls" is just wrong! The basis for including the 3830-2/3 in the director category is the undisputed fact that the 3333-11, 3340-A2 and 3350-A2 all attach without modification to the directors of the 3880, attach without modifcation to the 3830-2 and do not attach to the 3830-1. The 3880 and 3090 documents establish that this interface is the CTL-I. There are maintenance documents, probably at the Computer History Museum or IBM archives that will further confirm that the host side interfaces of all the A-Units are essentially the same CTL-I, ditto for the drive side of all three director type storage controls. These are all reliable sources. There is also no dispute that the interface on the channel side is also the same thing. So it is logical and not synthesis to categorize the 3830-2/3 under the director type and not under the conventional type. Just because u can't find the word director in some of the 3830-2/3 documents is not proof that they are not "Director type storage controls;" and is not a justification for the massive edit. Therefore I reluctantly (not capriciously) reverted.

You should note that I did not revert your 3380-3/3350 edit even though to the best of my recollection it is incorrect.

I left your {{Cn}} on the 3380s in the Director type section to see if some other editor might comment. If you make mistakes I will correct if possible and only revert when it just is easier as is this case. Please stop the ad hominem attacks Tom94022 (talk) 06:49, 25 March 2016 (UTC)

PKB. Your "You have a propensity" is also ad hominem.

From http://www.ibm.com/software/globalization/terminology/s.html

storage director
A physical or logical element that manages multiple paths to storage components, such as devices or controllers. See also storage path.

From http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/A13I1005/GLOSSARY?SHELF=EZ2HW125&DT=19960917232359#SPTGLOSS

storage director. In a 3990 storage control, a logical entity consisting of one or more physical storage paths in the same storage cluster. In a 3880, a storage director is equivalent to a storage path. See also storage path, single-path storage director, and multipath storage director.
In a 3990 storage control, a logical entity consisting of one or more physical storage paths in the same storage cluster. In a 3880, a storage director is equivalent to a storage path. See also storage path, single-path storage director, and multipath storage director.

Also note that the 3880 and 3990 manual show the directors as components of the controller. Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:49, 25 March 2016 (UTC)

Nothing quoted directly above is particularly responsive to the analysis immediately above the quoted material. The first quote from a software perspective is not a reliable source in this matter. Nothing about the second quote contradicts the categorization discussion above and now below. At the risk of somewhat repeating the above and I don't have much time to go thru it in detail with links right now, but it goes something like this.

• The best definition of director, controller, A-Unit and B-Unit is to the best of my recollection in the 3990 ref manual glossary. The manual also clearly establishes the interface between the director and A-Unit is CTL-I (two fat channel cables with very different control signalling as i recall) It supports 3380s & 3390s on its directors
• The 3880 ref manual shows the interface as CTL-I and any of its directors can support 3380, 3350, 3340 and 3330 A-Units One 3880 verision has a single director, most directors have two directors as components but it seems sematically correct to say a storage control can have one director as its component - as in the 3880-4 (I think) and therefore also by extension in the 3830-2/3.
• Since the same 3330, 3340 and 3350 A-Units also attach to the 3830-2/3 using the CTL-I and not to the 3830-1 (which has a different interface) the former must be in a different category and given the common interfaces to the known directors and A-Units it is logical to so categorize them that way.

If the 3830-2/3 are not properly typed as directors, how shoud they be categorized? Tom94022 (talk) 08:48, 26 March 2016 (UTC)

PS: If propensity is an ad hominem it is yours, at least in this article's talk, and the only one in your last post. I appreciate the improvement and look forward to none. Tom94022 (talk) 08:56, 26 March 2016 (UTC)

As I've repatedly told you before, the 3880 manual explicitly refers to a director as a component of the control unit and, on pages 1-2 and 1-3, shows directors as limited to certain DASD types, e.g., director 1 of a 3880-2 cannot attach a string of 3380 DASD and director 2 of a 3880-2 cannot attach a string of 3330, 3340 or 3350 DASD.

The 3830-2 should just be categorized as a DASD controller. The 3830-3, as a component of the 3850 MSS is a more complex case, but I would probably still just call it a DASD controller. It's a tossup whether DASD controller or SCU reads better.

The joker in the pack is the 3350-C2/C2F; IBM changed the usage of the letter C with the 3380-CJ2. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:55, 30 March 2016 (UTC)

Yes the 3880 manual refers to director as a component; however the term storage director appears more than 300 times in the 3990 reference manual so it is not a term limited to the 3880. Therefore, I really don't see the relevance of your repeated comment to this discussion - did I miss something? The 3880 manual states "Through use of diskettes, each storage director can be initialized to attach the following types of disk storage devices" and lists all types; that is, any director can be intialized to attach any of the allowed sets of drives using a diskette. Since the initialization is with a diskette all the directors have the same CTL-I interface and all A-Units at the other end of the cables also have the same CTL-I interface!

It would be absolutely wrong to call the 3380-2 or -3 a DASD controller - the term "controller" is clearly associated with the conntrol funtion in an A-Unit (and C-Units too). I don't recall and don't have the time to research how the 3830-3 interconnected within the 3850, but as u pointed out it could attach A-Units in the form of 3350-A2s meaning it had to have a CTL-I and acted as a storage director to those units. Do you know the path by which records were moved from the 3851 to the 3830-2, I don't recall? It really doesn't matter though since it attached 3333s and 3350-A2s

You are correct in that the letter C has three meanings, an off line backup A-Unit, an online multipath A-Unit or a conventional storage control unit.

String Swithc is one other distinction not heretofore mentioned that further demonstrates the 3830-2 is a "director type." Architecturally SS was a disaster since it introduced a point of resourse contention that was not recognized by the OS. If a director attempted to connect to a device on a string currently in use by another director it faced a dilemena as to busy status. If it responded Device Busy it would imply the other up to 7 device on the string were available perhaps leading to thrashing as the OS attempted to access them. On the other hand if it responded Control Unit Busy it would deny availability of up to 24 drives on the other three strings. I forget what we did, I think it was Device Busy and then respond Device Busy for any other attemped access to that string. Didn't have that problem at all in conventional storage controls and not even in the 2844 since there were two paths to all drives. Multi-path controllers got rid of that problem.

This article groups storage contols into like categories to help the reader understand IBMs evolving CKD attachments. It uses consistently the last of a series of terms that evolved over twenty years. The fact that IBM never used the term "director" in any of its 3830-2 literature is not evidence that it is not of the director category as the term came to be used. The 3830-2 passes the duck test, if it looks like a director, interfaces to a host like a director and interfaces to DASD like a director then it is a director type and not a conventional storage control. Tom94022 (talk) 07:03, 31 March 2016 (UTC)

Please stop rebutting claims I never made. I never claimed that the use of the term director was limited to the 3880, and the 3990 manuals also refer to directors as components of the SCU.

The same page that says "Through use of diskettes, each storage director can be initialized to attach the following types of disk storage devices" lists limitations on which directors can be initialized for which devices. Look at the section labelled Device Configurations . Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:06, 31 March 2016 (UTC)

Well I am glad to see that we agree that the term director is generic.

I believe elsewhere it can be established that any director can be intialized to any supported configuration with diskettes. That is certainly how I read the one manual we have and nothing in the manual says it can't be done. So looking at the section labelled "Device Configuration" really doesn't change anything. Tom94022 (talk) 23:52, 31 March 2016 (UTC)

Then establish that there is an error in the cited documentation. The text on pages 1-2 and 1-3 clearly says that it cannot be done. Documentation trumps faith. Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:11, 1 April 2016 (UTC)

Actually the text on page 1-2 and 1-3 states that depending upon model either Director 1 or 2 or both can support the prior 3333, 3340 and 3350 A-Units. It also says Directors are initialized by a diskette. The manual says nothing about the differences between the Directors in different models other than the devices supported and says nothing about field conversion between models. My recollection is they were field convertable with diskettes and other hardware components not related to the CTL-I and I suppose u contend they were not.

But that really doesn't matter since the 3880 doc states Director 1 and/or Director 2 supports combinations of 3333, 3340 and 3350 A-Units and the same A-Units are supported without any change by the one control function within a 3830-2/3 it is reasonable to catergorize the 3830-2/3 control function as a type of director and include them within said category in this article. Tom94022 (talk) 00:25, 2 April 2016 (UTC)

Straw dummy; the issue is not whether you can configure a director to select one of the valid device types for that director. The issue is the devices supported, and "Device Configuration" is unambiguous. As to field conversion between models, if it were possible it would include replacement of hardware.

A related issue is that you seem to be violating Wikipedia:No original research in addition to ignoring what the cited documents say. Shmuel (Seymour J.) Metz Username:Chatul (talk) 21:28, 4 April 2016 (UTC)

Please provide a reference for your assertion that "the 3880 manual explicitly refers to a director as a component of the control unit." [empahsis added] I can find no such quote - I can hardly find the work component and certainly not in this context; what I can find is:

"The 3880 contains two storage directors. Each storage director operates independently so that each one provides the basic functions for storage control. The 3880 contains two storage directors. Each storage director operates independently so that each one provides the basic functions for storage control. That is, each storage director has its own data path, control path, and address for channel communication" (GA26-1661-3 & -9, p 1-1)

The later reference manual acknowledges a Storage Control with one Storage Director. Since the original 3880 only attached the same units as the 3380-2 it seems that I am simply applying IBM's more modern terminology to its prior 3830-2 which as u must admit, provides "basic functions for storage control. That is, each storage director [the 3830-2] has its own data path, control path, and address for channel communication." just like a director. Tom94022 (talk) 20:27, 6 April 2016 (UTC)

3O request in re use of invented term director type to describe IBM 3830-2 and 3830-3

The dispute is whether the IBM term director refers to a type of Storage Control Unit or only to a component of an SCU, and whether the 3830 models 2 and 3 should be categorized as director type SCUs. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:18, 6 April 2016 (UTC)

The dispute is whether it is acceptable to apply a new term retroactively. The term director did not come into IBM storage vernacular until well after the 3380-2 shipped, so the question is simply does the new term correctly applied to this prior art. At the announcement, the 3880 SCUs contained two directors; each 3380 director provided a separate path for up to four combinations of 3333, 3340-A2 or 3350-A2 units. A 3830-2 provides the same path to the identically same combinations of up to four 3333, 3340-A2 or 3350-A2 units that attach a director of the 3880; literally one could unplug the cable at a 3880 and plug it into a 3830-2 (or vice versa) and have the attached drives work. Subsequently, a 3880 storage control was announced having only one director. As it turned out, the only SCUs (including ISCs) that attach these devices are the 3830-2/3 and certain models of the 3880 so it seems reasonable to group them together and use the term "director" to categorize them.

It is misleading to say a director is a component of a 3830-2; there are no other components of an SCU relevant to this discussion. A 3880 has one or two directors. Tom94022 (talk) 20:27, 6 April 2016 (UTC)

Response to Third Opinion Request:

Disclaimers: I am responding to a third opinion request made at WP:3O. I have made no previous edits on History of IBM CKD Controllers and cannot recall any prior interaction with the editors involved in this discussion which might bias my response. The third opinion process (FAQ) is informal and I have no special powers or authority apart from being a fresh pair of eyes. Third opinions are not tiebreakers and should not be "counted" in determining whether or not consensus has been reached. My personal standards for issuing third opinions can be viewed here.

Opinion: One particularly wise Third Opinion Wikipedian, RegentsPark, once succinctly put the purpose of Third Opinions like this, "It's sort of like if you're having an argument on the street in front of City Hall and turn to a passer-by to ask 'hey, is it true that the Brooklyn Bridge is for sale?'." "I am simply applying IBM's more modern terminology to its prior 3830-2" is pretty clearly prohibited original research since it clearly takes fact A, analyzes its meaning, and applies it to fact B. However, I would also note that this dispute is probably not going to be solved by looking at manuals. Manuals are, by definition, primary sources published by the manufactuer and that policy says, "Do not analyze, evaluate, interpret, or synthesize material found in a primary source yourself; instead, refer to reliable secondary sources that do so." Primary sources can only be used to say exactly what they say. To take what it says in one manual and analyze it to try to figure out what it means beyond exactly what the words say violates that policy. To then compare what one manual says vs what another says and come up with a conclusion is also prohibited original research. Someone needs to find a secondary reliable source which settles this dispute. Until that happens, there's nothing to argue about and the material should be excluded from the article, remembering that the verifiability policy says in the BURDEN section that the obligation is on the editor introducing the material to provide a reliable source as defined by Wikipedia. Repeatedly restoring material without such a source after it has been challenged can be seen as disruptive editing and cause an editor to be sanctioned.

What's next: Once you've considered this opinion click here to see what happens next.—TransporterMan (TALK) 04:06, 7 April 2016 (UTC)

Thanks for taking the time to express your opinion and I don't dispute most of u wrote (other than the need for secondary sources), but I think your opinion is off point. There lots of OR and SYN on this Talk page, which is permitted, but none in the article. We are dealing with just the heading of two sections, currently:

2. Conventional storage controls

3. Director type storage controls

The two sections recognise the change in IBM SCU archictecture called the "New Attachment Strategy" which occured with the shipment of the 3830-2, the first item under Section 3. This is discussed without many references in the text following Section 3 and is well documented in reliable sources such as Oral History of Jack Harker, 3830-2 announcement and marketing literature and testimony in the US v IBM antitrust matter. I probably should add in the references but so far the lede is mostly not disputed. The essential change first introduced with the 3830-2 was to split the SCU into two pieces, moving the drive oriented electronics into the first drive of a string of drives. That is the case for all of the SCUs in Section 3. So there should be no dispute if the two sections were labeled:

2. Storage controls containing drive electronics

3. Storage controls without drive electronics

Which gets to the use of the phrase "Director type" in the heading tytle for Section 3. IBM's almost dictionary definition of a Storage Director is reproduced above so there is a reliable source for director as single path to a DASD controller, the latter being defined as "provides DASD path control and data-transfer functions". That is the term "Director" is used consistent with IBM's definition. The heading doesn't say the 3830-2 is a Director but just that the following SCUs all are of the same type. This is not OR but editorial license.

Personally I think the original categorization is more informative that tne second one. I totally reject the concept that the SCUs are not properly categorized. I suppose if you prefer the latter titles, I can live with it. I do see that I need to improve the lede of Section 2 and reference the lede of Section 3. Perhaps u might wait unitl I have done both before rendering an opinion on which title is best.

BTW, only IBM produces material to the depth of this article so we likely will mainly rely on such primary sources. Most of what is in the article does come from IBM sources. As I read wp:WPNOTRS these IBM sourses are reliable.

Again thanks for taking the time to voice an opinion. Tom94022 (talk) 19:08, 9 April 2016 (UTC)

The use of a special DASD model as an intermediate between the control unit and the other DASD models did not begin with the 3333, it began with the 2319-B1, so the dispute is more than the labeling of the sections. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:44, 31 May 2016 (UTC)

Scope of article

Is there any reason to not change This practice continued in IBM's larger mainframes thru System/370; to This practice continued in IBM's larger mainframes thru IBM Z;? Similarly, is there any reason to not include DASD subsystems more recent than the 3990/3390, e,g,. 2105, 9340 and 9345, 9394 RAMAC Array? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:23, 10 March 2024 (UTC)

I see no problem with the first proposed change. Adding the additional subsystems gets into the CKD emulating subsystems. A new section might be appropriate starting with IBM's 1994 RAMAC Array and probably including a mention of the STC Symmetrix. Tom94022 (talk) 17:32, 10 March 2024 (UTC)

the STC Symmetrix Presumably "the EMC Symmetrix". Guy Harris (talk) 19:17, 10 March 2024 (UTC)

I changed S/370 to IBM Z; I hope that people more familiar with the newer controllers can add details on them. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:38, 11 March 2024 (UTC)