|
Technical Report HW-MACS-TR-0098
Title |
Scalable Persistent Storage for Erlang Theory and Practice |
Authors |
Amir Ghaffari, Natalia Chechina, Phil Trinder |
Date |
2013-06-06 |
Abstract |
Distributed computing platforms, like clusters, grids or clouds, have become popular due to their scalability and cost-effectiveness. For taking the advantage of the scalable platforms, many programming languages have been designed to support distributed programming.
Erlang has become one of the most popular programming language for developing distributed application. As any other programming language, Erlang needs distributed databases for saving and retrieving data. In the RELEASE project, we are working to scale up Erlang/OTP for massively parallel and distributed platforms. In this paper we evaluate the scalability of some of the mostly used DBMSs among Erlang developers. Our study shows that in theory Riak is a scalable, available and failure-tolerant distributed database. We investigate the scalability limits of Riak in practice, and find that Riak doesn't scale beyond ˜60 nodes. Moreover, our benchmark reveal that Riak provides a highly available and fault tolerant service.
|
Group |
DSG |
Notes |
|
Download |
|
|
|