You are viewing a plain text version of this content. The canonical link for it is here.

Posted to hdfs-user@hadoop.apache.org by "Mesika, Asaf" <as...@gmail.com> on 2012/11/13 01:21:49 UTC

Counters implementation

Hi,

Can anyone reference me to an article/blog detailing how Counters are implemented in HBase?

We get very low throughput of batch of Increment relative to batch of Put, and I would like to investigate why, by understanding the basic workflow on updating a counter - does it use block-cache? Does it access disk (IO) on each Increment (to read latest value) ?

Tnx,

Asaf

Re: Counters implementation

Posted by Ted Yu <yu...@gmail.com>.

I found the following:

http://cloudfront.blogspot.com/2012/06/hbase-counters-part-i.html
http://palominodb.com/blog/2012/08/24/distributed-counter-performance-hbase-part-1

Which hbase version are you using ?

Cheers

On Mon, Nov 12, 2012 at 4:21 PM, Mesika, Asaf <as...@gmail.com> wrote:

> Hi,
>
> Can anyone reference me to an article/blog detailing how Counters are
> implemented in HBase?
>
> We get very low throughput of batch of Increment relative to batch of Put,
> and I would like to investigate why, by understanding the basic workflow on
> updating a counter - does it use block-cache? Does it access disk (IO) on
> each Increment (to read latest value) ?
>
> Tnx,
>
> Asaf
>
>

Re: Counters implementation

Posted by Ted Yu <yu...@gmail.com>.

I found the following:

http://cloudfront.blogspot.com/2012/06/hbase-counters-part-i.html
http://palominodb.com/blog/2012/08/24/distributed-counter-performance-hbase-part-1

Which hbase version are you using ?

Cheers

On Mon, Nov 12, 2012 at 4:21 PM, Mesika, Asaf <as...@gmail.com> wrote:

> Hi,
>
> Can anyone reference me to an article/blog detailing how Counters are
> implemented in HBase?
>
> We get very low throughput of batch of Increment relative to batch of Put,
> and I would like to investigate why, by understanding the basic workflow on
> updating a counter - does it use block-cache? Does it access disk (IO) on
> each Increment (to read latest value) ?
>
> Tnx,
>
> Asaf
>
>

Re: Counters implementation

Posted by Ted Yu <yu...@gmail.com>.

I found the following:

http://cloudfront.blogspot.com/2012/06/hbase-counters-part-i.html
http://palominodb.com/blog/2012/08/24/distributed-counter-performance-hbase-part-1

Which hbase version are you using ?

Cheers

On Mon, Nov 12, 2012 at 4:21 PM, Mesika, Asaf <as...@gmail.com> wrote:

> Hi,
>
> Can anyone reference me to an article/blog detailing how Counters are
> implemented in HBase?
>
> We get very low throughput of batch of Increment relative to batch of Put,
> and I would like to investigate why, by understanding the basic workflow on
> updating a counter - does it use block-cache? Does it access disk (IO) on
> each Increment (to read latest value) ?
>
> Tnx,
>
> Asaf
>
>

Re: Counters implementation

Posted by Ted Yu <yu...@gmail.com>.

I found the following:

http://cloudfront.blogspot.com/2012/06/hbase-counters-part-i.html
http://palominodb.com/blog/2012/08/24/distributed-counter-performance-hbase-part-1

Which hbase version are you using ?

Cheers

On Mon, Nov 12, 2012 at 4:21 PM, Mesika, Asaf <as...@gmail.com> wrote:

> Hi,
>
> Can anyone reference me to an article/blog detailing how Counters are
> implemented in HBase?
>
> We get very low throughput of batch of Increment relative to batch of Put,
> and I would like to investigate why, by understanding the basic workflow on
> updating a counter - does it use block-cache? Does it access disk (IO) on
> each Increment (to read latest value) ?
>
> Tnx,
>
> Asaf
>
>