[
https://issues.apache.org/jira/browse/CASSANDRA-6477?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14619451#comment-14619451
]
Alan Boudreault commented on CASSANDRA-6477:
--------------------------------------------
This morning, I did some initial performance testing to get a first overview. I
tested the following scenarios:
* Writing without MV
* Writing with 1 MV
* Writing with 3 MV
* Writing with 5 MV
Results:
http://riptano.github.io/cassandra_performance/graph_v5/graph.html?stats=cassandra-mv-benchmark-2015-07-08.log&metric=op_rate&operation=1_write&smoothing=1&show_aggregates=true&xmin=0&xmax=341.77&ymin=0&ymax=89079.1
Briefly, with 1 MV created, which produces more mutations internally.. we got
33% of the initial performance. Surprisingly, I don't see any performance
decrease with 3 and 5 MVs as I would have expected. This leads us to think that
it might be caused by bottleneck (CPU? IO? some internal locks?)
My next step is to profile what's going on to understand more these results.
> Materialized Views (was: 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 beta 1
>
> Attachments: test-view-data.sh
>
>
> 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)
