[
https://issues.apache.org/jira/browse/CASSANDRA-5062?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13588834#comment-13588834
]
Jonathan Ellis commented on CASSANDRA-5062:
-------------------------------------------
I think I can summarize things this way:
# Embedding or requiring ZK is a non-starter. So anything that starts with
that as a dependency (Spinnaker or Kafka-style replication) means "implement
ZAB [or Paxos] first."
# A global master to designate cohort leaders as in HBase or Hibari is also a
non-starter.
# ZAB recovery after leader failure is a mess. (I recommend the 2008 paper
over the 2011 one; it's far more readable.)
# All of the options require some kind of "proposal storage" separate from
committed rows.
So I keep coming back to, "first implement raw Paxos; then we can use that at a
building block to optimize later if CAS performance becomes more of a
priority." (Proposal storage may in fact be the initial bottleneck, not the
replication strategy.)
> 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
