Cassandra vs. Voldemort benchmark

[Kristian Eide <kr...@gmail.com>]

There is a benchmark comparing Cassandra to Voldemort performance here:

http://blog.medallia.com/2010/05/choosing_a_keyvalue_storage_sy.html

-- 
Kristian

Re: Cassandra vs. Voldemort benchmark

[Benjamin Black <b...@b3k.us>]

My comment on this post:

"This is an interesting start to performance testing these systems,
but raises many more questions than it answers. I am disappointed you
chose not to investigate the enormous, unexplained spreads in
performance for either system tested, nor to attempt to adjust tuning
parameters to improve any of the metrics reported."

On Fri, May 7, 2010 at 8:09 PM, Kristian Eide <kr...@gmail.com> wrote:
> There is a benchmark comparing Cassandra to Voldemort performance here:
>
> http://blog.medallia.com/2010/05/choosing_a_keyvalue_storage_sy.html
>
> --
> Kristian
>

RE: Cassandra vs. Voldemort benchmark

[Todd Burruss <bb...@real.com>]

i did a lot of comparisons between voldemort and cassandra and in the end i decided to go with cassandra.  the main reason was recovery and balancing operations.  on the surface voldemort is s*** hot fast, until you need to restore a node or add nodes.  BDB (the default persistence solution) isn't very good at iterating over the entire key space.  it creates a _lot_ of random seeking across the disk and is therefore very slow.  for small data sets voldemort is a winner as you can have all your data cached.

(of course i'm not even considering data modeling, just performance on get/put/delete)

i do agree that voldemort and cassandra are first class projects and worthy of many production deployments.  right tool for the job.

________________________________________
From: Kristian Eide [kreide@gmail.com]
Sent: Friday, May 07, 2010 8:09 PM
To: user@cassandra.apache.org
Subject: Cassandra vs. Voldemort benchmark

There is a benchmark comparing Cassandra to Voldemort performance here:

http://blog.medallia.com/2010/05/choosing_a_keyvalue_storage_sy.html

--
Kristian