[Poster] A Stress Test of Titan (6 machine cluster - 14 machine read/write servers)

[Marko Rodriguez]

Hello,

Matthias Broecheler and I are at GraphLab 2012 presenting a poster about Titan.

http://thinkaurelius.github.com/titan/doc/titan-stress-poster.pdf

This poster presents an overview of Titan along with some excellent stress testing done by Matthias and Dan LaRoque. The stress test uses a 6 machine Titan cluster with 14 read/write servers slamming Titan with various read/writes. The results are presented in terms of the number of bytes being read/write from disk, the average runtime of the queries, the cost of a transaction on Amazon EC2, and a speculation of the number of concurrent users are concurrently interacting.

Enjoy!,

Marko.

http://markorodriguez.com

[Daniel Quest]

Awesome!  I am very impressed by all the colors wish I could be there!  

-Dan

[Matthias Broecheler]

No, we have not yet done an extensive Titan over HBase evaluation. If somebody has a particular use case they would like to evaluate the HBase storage backend for, please let us know.

On Tue, Jul 10, 2012 at 10:46 PM, McKinley <mckinl...@gmail.com> wrote:

Do you have plans to test and calculate the pricing HBase too?

--
Matthias Broecheler, PhD
http://www.matthiasb.com
E-Mail: m...@matthiasb.com