cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michaël Figuière (JIRA) <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-6477) Global indexes
Date Thu, 18 Dec 2014 22:02:16 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-6477?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14252369#comment-14252369
] 

Michaël Figuière commented on CASSANDRA-6477:
---------------------------------------------

Something that hasn't been mentioned so far in this thread is client side's Token Aware Balancing.
So far when a query relies on a Secondary Index, it doesn't have any partition key specified
in its where clause which means that the driver will fall back on the underlying balancing
policy, which is DC aware round robin by default in the DS Java Driver. That's the appropriate
behavior as there's no node that can be better than another as a coordinator in this situation.

With Global Indexes, in order for the Driver to still be able to perform Token Aware Balancing,
it'll need to be able to figure out which index will be used, which doesn't seems to be always
trivial in the above examples, especially if several columns and/or indexes are involved in
the {{WHERE}} clause. So here we might need to include an extra information about it in the
Prepare response message of the Native Protocol, if the indexes to be involved can be figured
out at Prepare-time.

> Global indexes
> --------------
>
>                 Key: CASSANDRA-6477
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-6477
>             Project: Cassandra
>          Issue Type: New Feature
>          Components: API, Core
>            Reporter: Jonathan Ellis
>            Assignee: Carl Yeksigian
>              Labels: cql
>             Fix For: 3.0
>
>
> Local indexes are suitable for low-cardinality data, where spreading the index across
the cluster is a Good Thing.  However, for high-cardinality data, local indexes require querying
most nodes in the cluster even if only a handful of rows is returned.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message