Geek Answers Handbook

Riac vs Gloucesterf

I need to set up a data store that can store the level of PB files (files are mostly small json, images and csv files, but some of them may be ~ 100 MB of binary files).

I am studying a distributed data warehouse that is non-mastering and does not have a single point of failure.

And I found Riak and GlusterFS.

I want to ask if any of you have used both of them before?

I know that the interface (DB / Map) is very different. But it seems to me that they both use hashing and similar distributed technologies. Will they have similar performance, consistency, and availability?

+8
filesystems cluster-computing riak
source share
3 answers

We are launching a Riak cluster with a size of 17 GB (24 GB of RAM, 2T) with a back-up of Bitkask, which stores about 1 billion 3 thousand objects. This setup is efficient, but very resource intensive. We are considering moving from Riak to GlusterFS, because performance is not so important to us. Perhaps using LevelDB as a backend will also reduce our problems.

Riak’s ATM self-healing properties seem stronger, and the configuration seems a bit simpler. In your case, I will be more comfortable storing 100 MB of files in GlusterFS.

+4
source share

Saving large files, such as the 100MB files you mentioned, would not be the right choice for a simple OiS Riak.

In this case, you should use the recently announced RiakCS http://basho.com/products/riakcs/ from Basho.

0
source share

The choice depends mainly on the requirements. In general, I would recommend Riak if you really do not need a real file system (with mounting points, ACL management, etc.) And it will just use or maintain files programmatically, and GlusterFS otherwise.

0
source share

More articles:


All Articles