[
https://issues.apache.org/jira/browse/CASSANDRA-2013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12994223#comment-12994223
]
Narendra Sharma commented on CASSANDRA-2013:
--------------------------------------------
We are using Cassandra as storage to build a message store. There are few
customers who are asking for two copies of data even when two replicas are
down. These are big customers with large number of subscribers and large data
volume. The most obvious solution we recommended is RF = 5 and R/W CL = QUORUM.
However, this increases the TCO. Keeping 5 copies of data makes the system less
attractive. Based on logical reason that you cannot have two working copies
unless you have two working replicas, they agreed to have RF = 4.
To achieve the requirement explained above with RF=4 we will need to start with
R/W CL = QUORUM. If one of the replica goes down we continue with CL=QUORUM (
we can downgrade to TWO as well). If second replica goes down, we reduce the CL
to TWO. This way we satisfy the R + W > N while two replica's are down, also
meet the requirement of maintaining two working copies even after two replicas
are down. As the replicas come up we upgrade the CL to QUORUM. We are aware
that this strategy has some loose ends where we may have partitioned reads and
writes with CL.TWO.
We believe this is a common usecase and other users of Cassandra will also find
this useful. Hence, atleast CL.TWO make lot of sense in practical cases.
> Add CL.TWO, CL.THREE; tweak CL documentation
> --------------------------------------------
>
> Key: CASSANDRA-2013
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2013
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Peter Schuller
> Assignee: Peter Schuller
> Priority: Minor
> Fix For: 0.8
>
> Attachments: 2013.txt
>
>
> Attaching draft patch to add CL.TWO and CL.THREE.
> Motivation for adding is that having to select between either ONE or QUORUM
> is too narrow a choice for clusters with RF > 3. In such a case, it makes
> particular sense to want to do writes at e.g. CL.TWO for durability purposes
> even though you are not looking to get strong consistency with QUORUM.
> CL.THREE is the same argument. TWO and THREE felt reasonable; there is no
> objective reason why stopping at THREE is the obvious choice.
> Technically one would want to specify an arbitrary number, but that is a much
> more significant change.
> Two open questions:
> (1) I adjusted the documentation of ConsistencyLevel to be more consistent
> and also to reflect what I believe to be reality (for example, as far as I
> can tell QUORUM doesn't send requests to all nodes as claimed in the .thrift
> file). I'm not terribly confident that I have not missed something though.
> (2) There is at least one unresolved issue, which is this assertion check
> WriteResponseHandler:
> assert 1 <= blockFor && blockFor <= 2 *
> Table.open(table).getReplicationStrategy().getReplicationFactor()
> : String.format("invalid response count %d for replication factor
> %d",
> blockFor,
> Table.open(table).getReplicationStrategy().getReplicationFactor());
> At THREE, this causes an assertion failure on keyspace with RF=1. I would, as
> a user, expect UnavailableException. However I am uncertain as to what to do
> about this assertion. I think this highlights one TWO/THREE are different
> from previously existing CL:s, in that they essentially hard-code replicate
> counts rather than expressing them in terms that can by definition be served
> by the cluster at any RF.
> Given that with THREE (and not TWO, but that is only due to the
> implementation detail that bootstrapping is involved) implies a replicate
> count that is independent of the replication factor, there is essentially a
> new failure mode. It is suddenly possible for a consistency level to be
> fundamentally incompatible with the RF. My gut reaction is to want
> UnavailableException still, and that the assertion check can essentially be
> removed (other than the <= 1 part).
> If a different failure mode is desired, presumably it would not be an
> assertion failure (which should indicate a Cassandra bug). Maybe
> UnstisfiableConsistencyLevel? I propose just adjusting the assertion (which
> has no equivalent in ReadCallback btw); giving a friendlier error message in
> case of a CL/RF mismatch would be good, but doesn't feel worth introducing
> extra complexity to deal with it.
> 'ant test' passes. I have tested w/ py_stress with a three-node cluster and
> an RF=3 keyspace and with 1 and 2 nodes down, and get expected behavior
> (available or unavailable as a function of nodes that are up).
--
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
