We see things more in the light of Heidi Howard’s recent Flexible Quorums and her comments that all these ideas can be mixed and matched.
Further, our reasons for not picking Raft are different, and have more to do with “what makes Raft Raft”, namely the concessions around view change and replacing message passing with RPC.
We’ve spoken a bit on this if you’d like to dive deeper:
Nit! It’s a bit more historically accurate to say that MultiPaxos/Raft are later variants of VSR since Brian Oki’s VSR predated Paxos by a year (‘88 vs ‘89) and Liskov and Cowling’s revision of Brian’s work in 2012 predated Raft by two years (the papers are remarkably similar, but Raft makes concessions for the sake of presentation).
I know it was published first, we’ve talked about this before :)
But, I’m not sure what was published first decide what’s a variant of what. I would say that, given the breadth of research into variants of Paxos and the ways it can be modified, it is most meaningful today to say they’re all variants of Paxos.
VSR having had little to no research or industry application until recently has a pretty weak claim. It does not appear to have influenced either Paxos or Raft. Raft was influenced by Paxos, and even VSR revisited discusses it in relation to these protocols.
In fact, the Raft paper cites that it was most influenced by VSR:
“Raft is similar in many ways to existing consensus algorithms (most notably, Oki and Liskov’s Viewstamped
Replication [29, 22])”
Happy to keep having this conversation, if only to shine a spotlight and pay tribute to some of the (lesser known but nevertheless) pioneers of our field. :)
I don’t interpret those words that way. I see that as a recognition of the VSR paper, as had been recently highlighted in VSR revisited at the time of publication. I guess you would have to ask the author if VSR had actually influenced his work, it’s certainly possible, but not the inference I would make from that snippet.
The paper references Paxos something like 100 times, versus 3 for VSR. It defines itself as a more understandable alternative to Paxos, so it was certainly influenced both by the existence and relevance of Paxos, and also in opposition to its apparent difficulty.
A good example to illustrate this perhaps is Babbage. He invented the computer first, but nobody using computers today was influenced by him, impressive though his achievements were! Nor would we say that computers are a kind of Babbage “analytical engine”. We say they are a kind of computer.
Diego there is only referring to “the VRR paper” (note the double “R”), i.e. specifically the VR “Revisited” paper of Cowling and Liskov in 2012 (not Oki and Liskov’s ‘88 work, which has a different title).
I wish I could share with you some of the anecdotes I’ve been privy to, having dived into the events and personally interviewed some of the people involved.
The history (or total order!) of consensus is fascinating here, almost like a Greek island, but only a few people will ever know it.
There were some great talks. The chat was satisfying. I like Zulip and have some experience with it so it was easy to navigate for me. It does take some getting used to though; a tips or FAQ would probably help a lot of folks.
An idea to improve your sanity re planning: Communicate a deadline of talks two weeks before the event and accept nothing 7 days after that.
Some people need hard deadlines to get their work done.
the reason I didn't attend / pay this time btw was because last time the talks were more about the "user-end of databases" and I mostly work in product and design; this time it was about more low-level stuff which is out of my scope. It's still cool stuff! But just outside of my focus / realm
charge more default $128, allow 64 and 32 then automated form for $10 “student tier” ask for documentation and auto approve. layer in twitter ads, linkedin, id love to see you make $100k per conf why not? don’t you think the market is there? in fact linkedin is your channel and why not start at $256, your conf is world class! companies will pay!
You can either do it through the Stripe dashboard by creating a Payment Link with a "Customer chooses what to pay" type or through the API (see https://docs.stripe.com/payment-links/api).
Worth mentioning that all the talks are published at https://www.hytradboi.com/2025/#program.
My favorite talk (although I haven't gotten through them all yet) was Alex Miller's Enough with all the Raft https://www.hytradboi.com/2025/2016d6c4-b08d-40b3-af2f-67926...
I wonder how TigerBeetle fits in this mental framework for OLTP replication.
We see things more in the light of Heidi Howard’s recent Flexible Quorums and her comments that all these ideas can be mixed and matched.
Further, our reasons for not picking Raft are different, and have more to do with “what makes Raft Raft”, namely the concessions around view change and replacing message passing with RPC.
We’ve spoken a bit on this if you’d like to dive deeper:
https://m.youtube.com/watch?v=tRgvaqpQPwE
TigerBeetle uses VSR, which is basically a variant of MultiPaxos/Raft.
Nit! It’s a bit more historically accurate to say that MultiPaxos/Raft are later variants of VSR since Brian Oki’s VSR predated Paxos by a year (‘88 vs ‘89) and Liskov and Cowling’s revision of Brian’s work in 2012 predated Raft by two years (the papers are remarkably similar, but Raft makes concessions for the sake of presentation).
I know it was published first, we’ve talked about this before :)
But, I’m not sure what was published first decide what’s a variant of what. I would say that, given the breadth of research into variants of Paxos and the ways it can be modified, it is most meaningful today to say they’re all variants of Paxos.
VSR having had little to no research or industry application until recently has a pretty weak claim. It does not appear to have influenced either Paxos or Raft. Raft was influenced by Paxos, and even VSR revisited discusses it in relation to these protocols.
In fact, the Raft paper cites that it was most influenced by VSR:
“Raft is similar in many ways to existing consensus algorithms (most notably, Oki and Liskov’s Viewstamped Replication [29, 22])”
Happy to keep having this conversation, if only to shine a spotlight and pay tribute to some of the (lesser known but nevertheless) pioneers of our field. :)
I don’t interpret those words that way. I see that as a recognition of the VSR paper, as had been recently highlighted in VSR revisited at the time of publication. I guess you would have to ask the author if VSR had actually influenced his work, it’s certainly possible, but not the inference I would make from that snippet.
The paper references Paxos something like 100 times, versus 3 for VSR. It defines itself as a more understandable alternative to Paxos, so it was certainly influenced both by the existence and relevance of Paxos, and also in opposition to its apparent difficulty.
A good example to illustrate this perhaps is Babbage. He invented the computer first, but nobody using computers today was influenced by him, impressive though his achievements were! Nor would we say that computers are a kind of Babbage “analytical engine”. We say they are a kind of computer.
Ha, as it happens there's documentary evidence online from Diego himself, that he was not influenced by VSR.
https://groups.google.com/g/raft-dev/c/cBNLTZT2q8o
Diego there is only referring to “the VRR paper” (note the double “R”), i.e. specifically the VR “Revisited” paper of Cowling and Liskov in 2012 (not Oki and Liskov’s ‘88 work, which has a different title).
I wish I could share with you some of the anecdotes I’ve been privy to, having dived into the events and personally interviewed some of the people involved.
The history (or total order!) of consensus is fascinating here, almost like a Greek island, but only a few people will ever know it.
Fair enough!
There were some great talks. The chat was satisfying. I like Zulip and have some experience with it so it was easy to navigate for me. It does take some getting used to though; a tips or FAQ would probably help a lot of folks.
The concept and branding are great.
If another one ends up happening I’d do it again.
An idea to improve your sanity re planning: Communicate a deadline of talks two weeks before the event and accept nothing 7 days after that. Some people need hard deadlines to get their work done.
the reason I didn't attend / pay this time btw was because last time the talks were more about the "user-end of databases" and I mostly work in product and design; this time it was about more low-level stuff which is out of my scope. It's still cool stuff! But just outside of my focus / realm
Feels strange to have posted exactly the same thing and gotten exactly zero traction.
Really glad this event exists. I didn't know about it ahead of time, so some amount of Marketing sounds sensible.
charge more default $128, allow 64 and 32 then automated form for $10 “student tier” ask for documentation and auto approve. layer in twitter ads, linkedin, id love to see you make $100k per conf why not? don’t you think the market is there? in fact linkedin is your channel and why not start at $256, your conf is world class! companies will pay!
> but now stripe supports pay-what-you-want directly (the docs say that they don't, but the docs are wrong)
Does anyone have links / suggestions on how to set this up?
You can either do it through the Stripe dashboard by creating a Payment Link with a "Customer chooses what to pay" type or through the API (see https://docs.stripe.com/payment-links/api).