[
https://issues.apache.org/jira/browse/CASSANDRA-5062?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13586321#comment-13586321
]
Jonathan Ellis commented on CASSANDRA-5062:
-------------------------------------------
bq. Paxos rounds get expensive with several parallel proposers (i.e., several
parallel user creations)
Of course. But what I'm proposing is basically paxos-per-replica-set, not a
single global paxos ensemble. So you get linearizability-per-partition (which
is the same as you'd get with a master-per-replica-set approach). As I said
above, I'm happy to accept "adequate" here. :)
bq. Complexity of the implementation apart
However, most of that complexity is because of the two "operation modes" -- as
you point out, leader election/recover, and 2PC. If we just stick with "raw"
paxos where any node may make a proposal, it's much simpler.
bq. why not using a simpler protocol, possibly lock-based
I'd be delighted, but as outlined above, I don't see how locks help avoid the
main problem around lost acks.
> Support CAS
> -----------
>
> Key: CASSANDRA-5062
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5062
> Project: Cassandra
> Issue Type: New Feature
> Components: API, Core
> Reporter: Jonathan Ellis
> Fix For: 2.0
>
>
> "Strong" consistency is not enough to prevent race conditions. The classic
> example is user account creation: we want to ensure usernames are unique, so
> we only want to signal account creation success if nobody else has created
> the account yet. But naive read-then-write allows clients to race and both
> think they have a green light to create.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
