Talk:Reliable multicast

	This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing articles
Low	This article has been rated as Low-importance on the project's importance scale.
	This article is supported by Networking task force.

Virtual synchrony was nominated for deletion. The discussion was closed on 16 January 2019 with a consensus to merge. Its contents were merged into Reliable multicast. The original page is now a redirect to this page. For the contribution history and old versions of the redirected article, please see its history; for its talk page, see here.

RMTP docs gone[edit]

The RMTP link provides no real information about RMTP. It does not seem to exist any more...217.18.21.2 (talk) 12:55, 5 July 2011 (UTC)[reply]

Streaming Media[edit]

I've removed the "or streaming media" example from the first paragraph, as streaming is a specific example of when packet retranmission does *not* help.

UDP or normal (unreliable) multicast is used for media streams, because late data is not useful. eg. it is of no use in an audio stream to request retransmission of dropped packets, as by the time they arrive the audio timeline will have moved on.

Granted, when using heavy buffering, the use of reliable multicast (or TCP in the case of unicast) *could* help, but this is very specific and still makes streaming a poor example.

Gladrim (talk) 13:48, 2 January 2018 (UTC)[reply]

Derecho[edit]

This wiki page will be merged with others as noted in the discussion on the proposal to delete. In that discussion, I mentioned that Derecho, a new Cornell system that combines virtual synchrony as a membership model with Paxos for doing updates to persistent data or for atomic multicast, would be out in 2019. So at this point Derecho is available on http://GitHub.com/Derecho-Project. The system is open source, free, and is fully described in an ACM Transactions on Computer Systems journal paper, available here: https://dl.acm.org/citation.cfm?id=3302258. I've added a mention of Derecho to the Ken Birman wiki page (someone else created one about my work, many years ago), and to the Paxos Protocol wiki page, on which many creators of Paxos protocols have described their variations. That particular page also was originally created many years ago by someone with no particular agenda, and these edits by others and myself have mostly (not entirely) been focused on the lists of systems using Paxos or the lists of Paxos protocols available for download.

The reason I mention all this is that Derecho is provably optimal for Paxos and virtual synchrony, so anyone who tackles the editing task to delete this page and merge it with others might want to point that out. Derecho is the first such optimal system, and actually the first open source library that merges virtual synchrony with Paxos in this way. The original idea of doing so was from a suggestion by Lamport and Malkhi, but they never took this further than a sketch of how one might replace the Paxos management model with a black box (they called it "Vertical Paxos") for tracking joins and leaves and for synchronizing those relative to updates and multicasts, using the virtual synchrony model for that aspect. So in this sense Vertical Paxos was a general template for "Paxos plus your favorite way of managing membership", and then virtual synchrony is one such way, and Derecho is a software library implementing all of this. This is just one aspect of Derecho, though -- it also has a storage (file system) layer, and a mapping to "RDMA" networks.

Anyhow, for whoever tackles that job, this might be useful context and perhaps worth citing or summarizing. Let me know if you have any questions.

Ken Birman (talk) 15:13, 26 May 2019 (UTC)[reply]