[HN Gopher] Paxos vs. Raft: Have we reached consensus on distrib...
       ___________________________________________________________________
        
       Paxos vs. Raft: Have we reached consensus on distributed consensus?
        
       Author : rbanffy
       Score  : 65 points
       Date   : 2020-04-27 09:57 UTC (13 hours ago)
        
 (HTM) web link (arxiv.org)
 (TXT) w3m dump (arxiv.org)
        
       | throw0101a wrote:
       | Ten minute video by co-author on this:
       | 
       | * https://www.youtube.com/watch?v=JQss0uQUc6o
       | 
       | See also her PhD dissertation, "Distributed consensus revised":
       | 
       | * https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-935.html
        
       | jganetsk wrote:
       | Heidi Howard is incredible and her work on distributed consensus
       | is illuminating. I think she has actually successfully cracked
       | the cookie of "making consensus easy".
       | https://www.youtube.com/watch?v=KTHOwgpMIiU
        
       | chapium wrote:
       | I thought raft was the obvious choice since it is a far simpler
       | framework.
        
         | cmckn wrote:
         | The conclusion of the paper is that there isn't actually a
         | significant difference between the algorithms. The Raft paper
         | is much clearer about implementation, but (as Heidi says) the
         | impl ideas from the Raft paper can be applied to Paxos in many
         | cases. Raft's leader election _is_ a bit more elegant and
         | results in a less complex implementation. The paper was a great
         | read!
        
       | enitihas wrote:
       | Is there any open source widely used implementation of multi
       | paxos, because single paxos doesn't seem useful by itself.
        
       | ccleve wrote:
       | The paper's main conclusion is accurate. Raft is more
       | understandable because of the clarity of the paper. But
       | implementation is very tricky. As I've written elsewhere it takes
       | weeks or months to write a solid implementation from scratch.
        
         | prismatk wrote:
         | Raft itself - rather than any framework in which you would
         | actually want to use it - is quite simple to implement. A few
         | classmates of mine and I implemented a barebones Raft instance
         | in about a weekend.
        
       | ideal0227 wrote:
       | I think all these kinds of papers are very confusing. Comparing
       | RSM (replicated state machine) to Paxos is just like comparing a
       | car to an engine. It makes very little or no sense.
       | 
       | In the original Paxos paper
       | (https://lamport.azurewebsites.net/pubs/paxos-simple.pdf), the
       | part 3 (RSM) is not extensively explained. There are countless
       | ways to use Paxos to implement RSM. Multipaxos/Raft/Epaxos try to
       | fill in that gap.
       | 
       | By any means, Paxos itself is 10x simpler than Raft or whatever.
       | Every time I heard a "distributed system" engineer said Paxos is
       | complicated, I know he/she does not have much experience in the
       | field or at least has never implemented the core consensus
       | part...
        
         | wahern wrote:
         | Indeed, in the paper they're comparing MultiPaxos to Raft.
         | 
         | EDIT: For others, here's a _very_ comprehensive (as of ~2018)
         | review of Paxos-related distributed consensus algorithms with
         | an exposition for each one: https://vadosware.io/post/paxosmon-
         | gotta-concensus-them-all/ That's 17 in all, excluding the
         | original Paxos paper. IMO, it should be linked anywhere Paxos
         | is discussed. The link has been posted twice before by others
         | on HN, but unfortunately hasn't seen any discussion, perhaps
         | because it speaks for itself.
        
       ___________________________________________________________________
       (page generated 2020-04-27 23:00 UTC)