3 ms·
Paper's network assumptions: "We assume a model of reliable communication—messages cannot be lost (only delayed). This is a reasonable assumption if we conside
by cioc 10y ago
Paper's network assumptions:
"We assume a model of reliable communication—messages cannot be lost (only delayed). This is a reasonable assumption if
we consider a reliable protocol, such as TCP."
Paxos network assumptions:
"Messages can take arbitrarily long to be delivered, can be duplicated, and can be lost, but they are not corrupted."
- http://research.microsoft.com/en-us/um/people/lamport/pubs/paxos-simple.pdf http://research.microsoft.com/en-us/um/people/lamport/pubs/p...
- mperham 10y agoSounds like Paxos could be implemented on top of UDP whereas AllConcur requires TCP?
- cioc 10y agoDuplicated and lost messages will happen in practice - TCP attempts to build reliable communication on an unreliable medium. Not clear from the paper if this matters. I think the larger misunderstanding of the paper is the requirement that paxos have a leader - lamport's synod algorithm does not require a leader, the leader is an optimization!
- grogers 10y agoWithout a long-lived leader, the performance would be so bad that I doubt there is a single production implementation that doesn't use that optimization.
- coldcode 10y agoNot really fair to compare to Paxos if the network assumption is different.