[
https://issues.apache.org/jira/browse/PHOENIX-1901?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15068713#comment-15068713
]
Mujtaba Chohan commented on PHOENIX-1901:
-----------------------------------------
h4. Write
Performance on the write side with transactions is ~2% slower when comparing to
non-transactional table.
!http://phoenix-bin.github.io/client/publish/write-transactions.png!
----
h4. Read
This [Pherf performance
run|http://phoenix-bin.github.io/client/publish/pherf-txn.html] compares
non-transactional vs transactional table query read time. Table was changed to
transactional by _alter table_ statement. Performance is negligibly slower for
some queries on transactional table.
http://phoenix-bin.github.io/client/publish/pherf-txn.html
----
todo for mutable data and indexes.
> Performance test transaction implementation for immutable data
> --------------------------------------------------------------
>
> Key: PHOENIX-1901
> URL: https://issues.apache.org/jira/browse/PHOENIX-1901
> Project: Phoenix
> Issue Type: Sub-task
> Reporter: James Taylor
> Assignee: Mujtaba Chohan
> Fix For: 4.7.0
>
>
> Placeholder for performance work to validate transaction implementation
> scales well
> - First set of tests can be to run regression suite with default set to make
> tables transactional (compared against standard run).
> - Second test would be to use Pherf for a standard platform workload.
> - Third test would be to use a workload with mutable data to see how well
> transactions scale.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
