[
https://issues.apache.org/jira/browse/HBASE-8370?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13694343#comment-13694343
]
Elliott Clark commented on HBASE-8370:
--------------------------------------
bq.this number is always 99 % for us on all clusters
That's why I said we need more decimal places for it.
bq.Also, the different b/w 82 % cache hit ratio to 99 % cache hit ratio is
enormous.
But that 82% doesn't tell you anything all by itself. For a given work load is
80% good or bad. You don't know. That percentage is really only useful if you
have a base line so it's equally informative uf the cache percentage to go from
99 and then falls to 98 or if it's 84 and falls to 83. Additionally gauges are
bad. They just don't tell a great story. There's a lot of lossy data there,
sampling times can skew your picture of what's actually happening. See
[~phobos182]'s slides (https://speakerdeck.com/phobos182/metrics-at-pinterest)
on why you should prefer counters over gauges.
That's why I said that derivative of cache miss count is the best way to look
at cache efficacy. It gives you an accurate count of the number of times you
have to go to hdfs (not really disk since there can be os cache there). It
also provides a good way to compare today to yesterday.
> Report data block cache hit rates apart from aggregate cache hit rates
> ----------------------------------------------------------------------
>
> Key: HBASE-8370
> URL: https://issues.apache.org/jira/browse/HBASE-8370
> Project: HBase
> Issue Type: Improvement
> Components: metrics
> Reporter: Varun Sharma
> Assignee: Varun Sharma
> Priority: Minor
>
> Attaching from mail to [email protected]
> I am wondering whether the HBase cachingHitRatio metrics that the region
> server UI shows, can get me a break down by data blocks. I always see this
> number to be very high and that could be exagerated by the fact that each
> lookup hits the index blocks and bloom filter blocks in the block cache
> before retrieving the data block. This could be artificially bloating up the
> cache hit ratio.
> Assuming the above is correct, do we already have a cache hit ratio for data
> blocks alone which is more obscure ? If not, my sense is that it would be
> pretty valuable to add one.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
