HBase in a virtual cluster

[Andrey <at...@gmx.net>]

Hi All,

Has anyone got experience in using HBase in a cluster of virtual machines (not
dedicated HW nodes)? Is that configuration reliable enough to build production
systems upon it?

As far as I understand, Java timing functions (like System.curentTimeMillis(),
Thread.sleep() etc.) work unprecisely in virtual machines, so any kind of
unsynchronization between nodes is possible.  For that reason (it’s my
understanding), the virtual clusters are generally bad idea when using HBase.
Correct me, if I’m wrong.

Any comments are appreciated.

RE: HBase in a virtual cluster

[Jonathan Gray <jg...@facebook.com>]

I'm not sure exactly what you're referring to with currentTimeMillis() being unreliable on virtual machines.

Regardless of your environment, you should be running NTP to synchronize clocks.

Otherwise, take a look in the mailing archives, there have been a number of lengthy discussions on HBase on EC2 (or other services).  Also check out the presentations on the wiki.  Andrew Purtell has a recent presentation on HBase+EC2.

JG

> -----Original Message-----
> From: Andrey [mailto:atimerbaev@gmx.net]
> Sent: Thursday, March 25, 2010 8:08 AM
> To: hbase-user@hadoop.apache.org
> Subject: HBase in a virtual cluster
> 
> Hi All,
> 
> Has anyone got experience in using HBase in a cluster of virtual
> machines (not
> dedicated HW nodes)? Is that configuration reliable enough to build
> production
> systems upon it?
> 
> As far as I understand, Java timing functions (like
> System.curentTimeMillis(),
> Thread.sleep() etc.) work unprecisely in virtual machines, so any kind
> of
> unsynchronization between nodes is possible.  For that reason (it’s my
> understanding), the virtual clusters are generally bad idea when using
> HBase.
> Correct me, if I’m wrong.
> 
> Any comments are appreciated.
>