News for hackers. Stuff that matters.
Re:Distributed systems theory for the distributed systems en... by einarvollset on Thursday May 12, 2016 @11:15PM (#11688408)
(Before you down vote: I have a PhD in distributed systems and fault tolerance. Okay, now you can down vote for the duchebaggery of this prescript)

I think a fundamental and very underrated paper and concept (which actually predates Paxos, yet Lamport ignored or was unaware of) is the notion of randomized consensus protocols. Simpler than "structured" leader type algorithms. Believe Ben Or's algorithm was first.

Re:Distributed systems theory for the distributed systems en... by mjb on Thursday May 12, 2016 @11:29PM (#11688450)
> Believe Ben Or's algorithm was first.

Ben-Or's "Another Advantage of Free Choice" beat Rabin's "Randomized Byzantine Generals" by a couple of months in 1983. These algorithms show how much people over-extend results like FLP. The result is about a very particular system model, and the addition of even a very tiny extra piece (in Ben-Or's case, a random oracle) makes the consensus problem possible again.

I wouldn't say that these algorithms were really ignored by Lamport when he wrote the Paxos paper. Again, they're solving a different problem in a different system model. If you want to pick on Lamport, talk about Liskov's Viewstamped Replication.

If anybody has a digital copy of Ben-Or's paper that isn't partially cut off, please make it available. Both the copy in the ACM library and the only copy the author himself has are missing some of the right hand side.

Re:Distributed systems theory for the distributed systems en... by einarvollset on Friday May 13, 2016 @11:31AM (#11690959)
I disagree - an ex-colleague at Cornell wrote a paper proving equivalence. Will have to dig that up..
Re:Distributed systems theory for the distributed systems en... by athenot on Thursday May 12, 2016 @11:19PM (#11688421)
An illustration is bees vs. flies.

A bee trapped indoors will go for the sun: a great algorithm for a forest/thicket but fatal in a house or car with windows. A fly will randmly try until it succeeds, meaning it is slower at escaping a thicket but will eventually find the open door even if it's opposite from the window.

Re:Distributed systems theory for the distributed systems en... by elviejo on Friday May 13, 2016 @06:41AM (#11689602)
Where did You study your degree? I would be intereses in doing one.