[
https://issues.apache.org/jira/browse/CASSANDRA-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14334211#comment-14334211
]
Russ Hatch commented on CASSANDRA-8692:
---------------------------------------
Initial benchmarks on ec2 c3.8xlarge nodes are complete. I ran this matrix as
suggested by [~enigmacurry]:
{noformat}
In memory load (fieldcount * fieldlength * recordcount * fluff < total memory
size):
readproportion: 1.0:
uniform
zipfian
latest
readproportion: 0.9, writeproportion: 0.10:
uniform
zipfian
latest
2x memory load (fieldcount * fieldlength * recordcount * fluff <= (2 * total
memory size)):
readproportion: 1.0:
uniform
zipfian
latest
readproportion: 0.9, writeproportion: 0.10:
uniform
zipfian
latest
{noformat}
As a control/baseline I used trunk as of late last week. The patch was from
https://github.com/aweisberg/cassandra/tree/C-8692-2 (git sha 422f5e194).
[in memory - read 100% - uniform
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_inmemory_uniform_read_100.json]
[in memory - read 100% - zipfian
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_inmemory_zipfian_read_100.json]
[in memory - read 100% - latest
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_inmemory_latest_read_100.json]
[in memory - read 90% write 10% - uniform
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_inmemory_uniform_read_90_write_10.json]
[in memory - read 90% write 10% - zipfian
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_inmemory_zipfian_read_90_write_10.json]
[in memory - read 90% write 10% - latest
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_inmemory_latest_read_90_write_10.json]
[2X memory - read 100% - uniform
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_2xmemory_uniform_read_100.json]
[2X memory - read 100% - zipfian
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_2xmemory_zipfian_read_100.json]
[2X memory - read 100% - latest
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_2xmemory_latest_read_100.json]
[2X memory - read 90% write 10% - uniform
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_2xmemory_uniform_read_90_write_10.json]
[2X memory - read 90% write 10% - zipfian
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_2xmemory_zipfian_read_90_write_10.json]
[2x memory - read 90% write 10% - latest
dist|http://riptano.github.io/cassandra_performance/ycsb/graph.html?stats=8692_2xmemory_latest_read_90_write_10.json]
> Coalesce intra-cluster network messages
> ---------------------------------------
>
> Key: CASSANDRA-8692
> URL: https://issues.apache.org/jira/browse/CASSANDRA-8692
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Ariel Weisberg
> Assignee: Ariel Weisberg
> Fix For: 2.1.4
>
> Attachments: batching-benchmark.png
>
>
> While researching CASSANDRA-8457 we found that it is effective and can be
> done without introducing additional latency at low concurrency/throughput.
> The patch from that was used and found to be useful in a real life scenario
> so I propose we implement this in 2.1 in addition to 3.0.
> The change set is a single file and is small enough to be reviewable.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
