[
https://issues.apache.org/jira/browse/PHOENIX-6211?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17261957#comment-17261957
]
Kadir OZDEMIR edited comment on PHOENIX-6211 at 1/9/21, 10:13 PM:
------------------------------------------------------------------
[~comnetwork], In order to disable it, we need to set
phoenix.server.page.size.ms to a large value. In my manual tests on a real
cluster over large tables, I did not see any performance difference for the
phoenix.server.page.size.ms values from 3 seconds to 60 seconds. I do not see
any reason to see performance impact if the page size is set to something in
seconds. The default value is set to half of the HBase RPC timeout value (i.e.,
typically the default value will be 30 seconds). There is no reason to expect
performance impact due to an extra RPC or restarting an HBase region scanner
per 30 seconds. Please note that most of the scan RPCs should complete in
milliseconds. The paging will happen only when the RPC takes more than 30
seconds.
was (Author: kozdemir):
[~comnetwork], In order to disable it, we need to set
phoenix.server.page.size.ms to a large value. In my manual tests on a real
cluster over large tables, I did not see any performance difference for the
phoenix.server.page.size.ms values from 3 seconds to 60 seconds. I do not see
any reason to see performance impact if the page size is set to something in
seconds. The default value is set to half of the HBase RPC timeout value (i.e.,
typically the default value will be 30 seconds). There is no reason to expect
performance impact due to an extra RPC or restarting a HBase region scanner per
30 seconds. Please note that most of the scan RPCs should complete in
milliseconds. The paging will happen only when the RPC takes more than 30
seconds.
> Paged scan filters
> ------------------
>
> Key: PHOENIX-6211
> URL: https://issues.apache.org/jira/browse/PHOENIX-6211
> Project: Phoenix
> Issue Type: Improvement
> Affects Versions: 5.0.0, 4.14.3
> Reporter: Kadir OZDEMIR
> Assignee: Kadir OZDEMIR
> Priority: Critical
> Fix For: 4.16.0
>
> Attachments: PHOENIX-6211.4.x.001.patch, PHOENIX-6211.4.x.002.patch,
> PHOENIX-6211.4.x.004.patch, PHOENIX-6211.4.x.005.patch,
> PHOENIX-6211.4.x.006.patch, PHOENIX-6211.master.001.patch,
> PHOENIX-6211.master.002.patch, PHOENIX-6211.master.003.patch,
> PHOENIX-6211.master.004.patch, PHOENIX-6211.master.005.patch,
> PHOENIX-6211.master.006.patch, PHOENIX-6211.master.007.patch,
> PHOENIX-6211.master.008.patch, PHOENIX-6211.master.009.patch,
> PHOENIX-6211.master.010.patch
>
>
> Phoenix performs two main operations on the server side: aggregation and
> filtering. However, currently there is no internal Phoenix paging capability,
> and thus server side operations can take long enough to lead to HBase client
> timeouts. PHOENIX-5998 and PHOENIX-6207 are for providing the paging
> capability for ungrouped and grouped aggregate operations. This improvement
> Jira is for adding the paging capability for scan filters.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
