[
https://issues.apache.org/jira/browse/IGNITE-5173?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Denis Magda updated IGNITE-5173:
--------------------------------
Description:
When the write-behind store reaches maximum write-behind queue/buffer size on
the next update operation, then the data will be flushed from the queue
synchronously in order to free some room for the just arrived update. The issue
is that the store flushes a single entry at a time in the synchronous mode that
can dramatically affect the performance.
The following has to be addressed as a part of the ticket:
* the store has to flush multiple entries in the sync mode to avoid a
performance hit.
* {{CacheConfiguration.getWriteBehindFlushSize}} and respective setter method
must be renamed to {{CacheConfiguration.getWriteBehindBufferSize}}. Presently
it's totally confusing.
See more details here:
http://apache-ignite-users.70518.x6.nabble.com/CacheStore-s-Performance-Drops-Dramatically-Why-td12337.html
was:
When the write-behind store reaches maximum write-behind queue/buffer size on
the next update operation, then the data will be flushed from the queue
synchronously in order to free some room for the just arrived update. The issue
is that the store flushes a single entry at a time in the synchronous mode that
can dramatically affect the performance.
The following has to be addressed as a part of the ticket:
* the store has to flush multiple entries in the sync mode to avoid a
performance hit.
* `CacheConfiguration.getWriteBehindFlushSize` and respective setter method
must be renamed to `CacheConfiguration.getWriteBehindBufferSize`. Presently
it's totally confusing.
See more details here:
http://apache-ignite-users.70518.x6.nabble.com/CacheStore-s-Performance-Drops-Dramatically-Why-td12337.html
> Write-Behind store flushes single entry at a time in the sync mode
> ------------------------------------------------------------------
>
> Key: IGNITE-5173
> URL: https://issues.apache.org/jira/browse/IGNITE-5173
> Project: Ignite
> Issue Type: Bug
> Reporter: Denis Magda
> Fix For: 2.1
>
>
> When the write-behind store reaches maximum write-behind queue/buffer size on
> the next update operation, then the data will be flushed from the queue
> synchronously in order to free some room for the just arrived update. The
> issue is that the store flushes a single entry at a time in the synchronous
> mode that can dramatically affect the performance.
> The following has to be addressed as a part of the ticket:
> * the store has to flush multiple entries in the sync mode to avoid a
> performance hit.
> * {{CacheConfiguration.getWriteBehindFlushSize}} and respective setter method
> must be renamed to {{CacheConfiguration.getWriteBehindBufferSize}}. Presently
> it's totally confusing.
> See more details here:
> http://apache-ignite-users.70518.x6.nabble.com/CacheStore-s-Performance-Drops-Dramatically-Why-td12337.html
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
