Hello,

I would like to get timing results for some large, rare scans, but each
reading test keeps performing better because of the heap's LRU block cache.
Is there a way (possibly through the API) that I can clear the block cache
to get timing results that mimic cold read times? I tried
HTable.clearRegionCache() but that didn't seem to do anything in terms of
making the scans taking longer.

Thanks.

Take a look at setCacheBlocks.

"Set whether blocks should be cached for this Scan.

This is true by default. When true, default settings of the table and
family are used (this will never override caching blocks if the block cache
is disabled for that family or entirely)"
http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html#setCacheBlocks(boolean)

On Thu, Jun 14, 2012 at 10:39 AM, Kevin <[EMAIL PROTECTED]> wrote:

> Hello,
>
> I would like to get timing results for some large, rare scans, but each
> reading test keeps performing better because of the heap's LRU block cache.
> Is there a way (possibly through the API) that I can clear the block cache
> to get timing results that mimic cold read times? I tried
> HTable.clearRegionCache() but that didn't seem to do anything in terms of
> making the scans taking longer.
>
> Thanks.
>

Kevin,

  One way to do it would be to use a configuration parameter and just set
it to zero for those scans.  From the HBase book:

 Disabling Blockcache

Do not turn off block cache (You'd do it by
setting hbase.block.cache.size to zero). Currently we do not do well if you
do this because the regionserver will spend all its time loading hfile
indices over and over again. If your working set it such that block cache
does you no good, at least size the block cache such that hfile indices
will stay up in the cache (you can get a rough idea on the size you need by
surveying regionserver UIs; you'll see index block size accounted near the
top of the webpage).

http://hbase.apache.org/book/important_configurations.html
On Thu, Jun 14, 2012 at 1:39 PM, Kevin <[EMAIL PROTECTED]> wrote:
>
> Hello,
>
> I would like to get timing results for some large, rare scans, but each
> reading test keeps performing better because of the heap's LRU block
cache.
> Is there a way (possibly through the API) that I can clear the block cache
> to get timing results that mimic cold read times? I tried
> HTable.clearRegionCache() but that didn't seem to do anything in terms of
> making the scans taking longer.
>
> Thanks.
--
Kevin O'Dell
Customer Operations Engineer, Cloudera