hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stack <st...@duboce.net>
Subject Re: Optimizations for random read performance
Date Thu, 18 Feb 2010 20:30:59 GMT
On Thu, Feb 18, 2010 at 7:52 AM, James Baldassari <james@dataxu.com> wrote:
> Here are the block cache hit ratio stats:
> Region server 1: 99
> Region server 2: 99
> Region server 3: 99
> Region server 4: 99
>

Looks like you are serving reads from cache.

> Sure.  Here's 'top | head' for all region servers:
> Region server 1: http://pastebin.com/m45e0bc2
> Region server 2: http://pastebin.com/m31868bb6
> Region server 3: http://pastebin.com/m5b1ce71f
> Region server 4: http://pastebin.com/mae7326f
>

Severly underloaded.  No 'wa' time with some decent CPU usage.  As you
say above, it like you have room to grow.

It would seem that you are still on 64k storefile blocksize (for the
reason you gave above) and yes, it doesn't sound like anything
interesting going on in your regionserver files.

St.Ack

Mime
View raw message