[
https://issues.apache.org/jira/browse/CASSANDRA-15066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16861604#comment-16861604
]
Joseph Lynch commented on CASSANDRA-15066:
------------------------------------------
[~vinaykumarcse] and I have been updating the patch authors and reviewers
continually on IRC and now ASF slack as we are running tests, but since this is
getting close to merge I just want to chime in that our small scale (12 node)
testing is showing excellent results so far. We've been working to validate
this patch from a real-world-deployment/scalability/performance perspective and
at this time the patchset appears more stable and more performant than 3.0. The
testing methodology and results are being recording in an open
[spreadsheet|https://docs.google.com/spreadsheets/d/1Vq_wC2q-rcG7UWim-t2leZZ4GgcuAjSREMFbG0QGy20/edit#gid=0]
that we are updating as we test and once this is merged we can start resuming
our formal tests as part of CASSANDRA-14746.
A summary of the results so far from our (Netflix) testing:
* The unbounded hints that we used to see under load on trunk are no longer
there.
* Better process level stability (thread CPU distribution, JVM allocation, etc
...)
* Excellent CPU flamegraphs and profiles, messaging is almost never the
dominant CPU factor.
* Performance appears superior in this patch to 3.0.x across the board
So far we have very thoroughly tested the following on the small cluster:
* LOCAL_ONE with variable read <-> write balance of 4kb multi column partitions
* LOCAL_QUORUM with variable read <-> write balance of 4kb multi-column
partitions
* QUORUM with variable read <-> write balance of 4kb multi-column partitions
We have also begun verification of the following combinations of messaging:
* Compression on, Encryption on
* Compression on, Encryption off
* Cross datacenter setups with 80ms of delay
> Improvements to Internode Messaging
> -----------------------------------
>
> Key: CASSANDRA-15066
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15066
> Project: Cassandra
> Issue Type: Improvement
> Components: Messaging/Internode
> Reporter: Benedict
> Assignee: Benedict
> Priority: High
> Fix For: 4.0
>
> Attachments: 20k_backfill.png, 60k_RPS.png,
> 60k_RPS_CPU_bottleneck.png, backfill_cass_perf_ft_msg_tst.svg,
> baseline_patch_vs_30x.png, increasing_reads_latency.png,
> many_reads_cass_perf_ft_msg_tst.svg
>
>
> CASSANDRA-8457 introduced asynchronous networking to internode messaging, but
> there have been several follow-up endeavours to improve some semantic issues.
> CASSANDRA-14503 and CASSANDRA-13630 are the latest such efforts, and were
> combined some months ago into a single overarching refactor of the original
> work, to address some of the issues that have been discovered. Given the
> criticality of this work to the project, we wanted to bring some more eyes to
> bear to ensure the release goes ahead smoothly. In doing so, we uncovered a
> number of issues with messaging, some of which long standing, that we felt
> needed to be addressed. This patch widens the scope of CASSANDRA-14503 and
> CASSANDRA-13630 in an effort to close the book on the messaging service, at
> least for the foreseeable future.
> The patch includes a number of clarifying refactors that touch outside of the
> {{net.async}} package, and a number of semantic changes to the {{net.async}}
> packages itself. We believe it clarifies the intent and behaviour of the
> code while improving system stability, which we will outline in comments
> below.
> https://github.com/belliottsmith/cassandra/tree/messaging-improvements
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
