Hbase on Amazon S3?

[Something Something <ma...@gmail.com>]

Anyone installed HBase on S3 (or EC2 for that matter)?  Any pointers would
be greatly appreciated.  Thanks.

Re: Hbase on Amazon S3?

[Andrew Purtell <ap...@apache.org>]

The new scripts in trunk at src/contrib/ec2 will offer this approach soon. Right now they simply back HDFS with instance storage (volatile) and rely on not having more than the HDFS replication factor (default = 3) instances crash or terminate at one time. Using EBS is a big win for its persistence and transparent/background snapshot facility. One thing our scripts will have to deal with though is how to back a ~100 or so node cluster with EBS volumes, and also supporting elastic operation, creating them on the fly as necessary. 

Also in the cards is performance and stability testing with HBase root filesystem on Hadoop's S3N fs (http://wiki.apache.org/hadoop/AmazonS3). I tried some limited testing with the S3 fs option just for basic filesystem operations -- albeit on a 209 GB file -- and had an unhappy result so will avoid that for now. Some time ago Clint Morgan ran a simple performance comparison and here was his results: http://markmail.org/message/xqhwgdw25oi7u3rb
"So to summarize:
loading data: almost twice as slow
A long scan is about 1.5 times slower
short scans are over an order of magnitude slower
and random reads (done on the sorted "scan") are over 2 orders of
magnitude slower"

In some fairly short time we should have a replacement for the HBase S3 related page up on the wiki. In the meantime you may consider perusing http://www.google.com/search?hl=en&q=hbase+s3

    - Andy




________________________________
From: Vaibhav Puranik <vp...@gmail.com>
To: hbase-user@hadoop.apache.org
Sent: Mon, November 16, 2009 9:46:52 AM
Subject: Re: Hbase on Amazon S3?

We have HBase 0.20.0 running on EC2 with EBS volume since July 2009.
We are using m1.Large machines for all the 4  nodes.

All of our data resides on EBS  volume. This helps us in backing up the
data. This also helps us in bringing up a separate cluster with the same
data for QA purposes.

So far no problems.

If you have any specific questions please let us know.

Regards,
Vaibhav Puranik
Gumgum



On Mon, Nov 16, 2009 at 9:38 AM, Something Something <
mailinglists19@gmail.com> wrote:

> Anyone installed HBase on S3 (or EC2 for that matter)?  Any pointers would
> be greatly appreciated.  Thanks.
>



      

Re: Hbase on Amazon S3?

[Vaibhav Puranik <vp...@gmail.com>]

We have HBase 0.20.0 running on EC2 with EBS volume since July 2009.
We are using m1.Large machines for all the 4  nodes.

All of our data resides on EBS  volume. This helps us in backing up the
data. This also helps us in bringing up a separate cluster with the same
data for QA purposes.

So far no problems.

If you have any specific questions please let us know.

Regards,
Vaibhav Puranik
Gumgum



On Mon, Nov 16, 2009 at 9:38 AM, Something Something <
mailinglists19@gmail.com> wrote:

> Anyone installed HBase on S3 (or EC2 for that matter)?  Any pointers would
> be greatly appreciated.  Thanks.
>

Re: Hbase on Amazon S3?

[stack <st...@duboce.net>]

A recent long thread herein was about hbase on s3.  In the end, there were
issues remaining (I believe -- someone correct me if I'm wrong please),
leaving aside the base fact that s3 is an eventually consistent filesystem
whereas hdfs wants to be strongly consistent -- at least when its running
under hbase.  The general opinion would seem to be avoid it; use ebs volumes
instead.

Regards s3, Andrew Purtell has just committed to trunk into contrib an ec2
directory.  It has scripts for generating ami's, etc.  Its derivative if the
contrib/ec2 that is up in our hadoop parent.  There are also some ami's that
andrew has published.  More on these soon....

St.Ack



On Mon, Nov 16, 2009 at 9:38 AM, Something Something <
mailinglists19@gmail.com> wrote:

> Anyone installed HBase on S3 (or EC2 for that matter)?  Any pointers would
> be greatly appreciated.  Thanks.
>