heavy reads

[shapoor <es...@yahoo.com>]

Hello,
i have an application which does a lot of sequential reads. I also save more
data, meaning the reads always have to be done in bigger amount of data in
storage (writes and reads don't happen parallel). 
To see how the hbase installation reacts to my needs, I evaluated the reads
and calculated out the min- max- average read time. There is often a huge
gap between max and mins and also the average does not stay contantly during
the process. Why does my hbase behave like this? should I change something
in the configuration?

regards,

forum.png <http://apache-hbase.679495.n3.nabble.com/file/n4056621/forum.png>  



--
View this message in context: http://apache-hbase.679495.n3.nabble.com/heavy-reads-tp4056621.html
Sent from the HBase User mailing list archive at Nabble.com.

Re: heavy reads

[shapoor <es...@yahoo.com>]

I read that pre-splitting the table is for write-maintenance. I haven't still
splitted. I have heavier writes than reads but as mentioned they don't occur
parallel. It's either read or write.

My application is running on a "single node" hbase-0.94.16.tar installation.

Thanks and regards,



--
View this message in context: http://apache-hbase.679495.n3.nabble.com/heavy-reads-tp4056621p4056695.html
Sent from the HBase User mailing list archive at Nabble.com.

Re: heavy reads

[Ted Yu <yu...@gmail.com>]

Have you pre-split your table ?
Are table regions evenly distributed across region servers ?

Which hbase release did you use ?

Thanks

On Mar 4, 2014, at 5:26 AM, shapoor <es...@yahoo.com> wrote:

> Hello,
> i have an application which does a lot of sequential reads. I also save more
> data, meaning the reads always have to be done in bigger amount of data in
> storage (writes and reads don't happen parallel). 
> To see how the hbase installation reacts to my needs, I evaluated the reads
> and calculated out the min- max- average read time. There is often a huge
> gap between max and mins and also the average does not stay contantly during
> the process. Why does my hbase behave like this? should I change something
> in the configuration?
> 
> regards,
> 
> forum.png <http://apache-hbase.679495.n3.nabble.com/file/n4056621/forum.png>  
> 
> 
> 
> --
> View this message in context: http://apache-hbase.679495.n3.nabble.com/heavy-reads-tp4056621.html
> Sent from the HBase User mailing list archive at Nabble.com.