hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Segel <michael_se...@hotmail.com>
Subject Re: hbase deployment using VMs for data nodes and SAN for data storage
Date Thu, 18 Oct 2012 08:39:56 GMT

I think we need to clarify what we think of as a  SAN. 
Its possible to have a SAN where the disks appear as attached storage, while the traditional
view is that the disks are detached. 

There are some design considerations like cluster density where one would want to use a SAN
like NetApp to effectively create a storage half to a cluster and then a compute half that
requires a fraction of the space and energy of a commodity built cluster. 

When we start to see clusters at PB scale, we have to consider the size of the footprint and
the cost of operating them in terms of both energy efficiency and physical footprint in a
data center. 

HBase can run in such configurations with the right tuning. 

I for one would love to have a data center where I can drop in different configurations and
be able to tune and validate cluster designs, but alas that's something only a MapR, Cloudera,
Hortonworks thing where they have the deep pockets and necessity to actually work through
this for their customers. 

On Oct 15, 2012, at 11:43 PM, lars hofhansl <lhofhansl@yahoo.com> wrote:

> If you have a SAN, why would you want to use HBase?
> -- Lars
> ________________________________
> From: "Pamecha, Abhishek" <apamecha@x.com>
> To: "user@hbase.apache.org" <user@hbase.apache.org> 
> Sent: Monday, October 15, 2012 3:00 PM
> Subject: hbase deployment using VMs for data nodes and SAN for data storage
> Hi
> We are deciding between using local disks for bare metal hosts Vs VMs using SAN for data
storage. I was wondering if anyone has contrasted performance, availability and scalability
between these two options?
> IMO, This is kinda similar to a typical  AWS or another cloud deployment.
> Thanks,
> Abhishek

View raw message