Distributed key-value stores

Posted in cloud, parallel by elisehuard on April 15, 2009

Distributed key-value stores are a hot topic of the moment.
It appears that there are many situations where the current relational database paradigm is not suited. Situations where simple key retrieval is enough – no joins or relations between tables.
And also distributed: this is more and more a requirement, and the traditional databases are not designed with distribution in mind, although there are various strategies in place to cope (sharding, duplications like master/slave setups).

This post is a good write-up of the more well-known ones. I’ve already posted about CouchDB.

Ilya Grigorik, always a good reference, is a big fan of the more obscure (well, over here at least) Tokyo Cabinet. I’m about to try this out in my current project, so I’ll post later on to say how it went.


2 Responses

Subscribe to comments with RSS.

  1. Steven Noels said, on April 15, 2009 at 7:49 pm

    Be very careful about the “distributed” character of CouchDB, I was bitten by the newspeak until I dug deeper. You can set up and trigger synchronisation, but it’s a manual operation. If you’re interested in ‘true’ P2P automatic replication, have a look at Cassandra, which sits underneath Facebook’s Inbox app.

  2. elisehuard said, on April 15, 2009 at 9:08 pm

    i agree with you, and so does the review i mention 🙂
    Cassandra gets criticised in that review as being not very well supported nor documented. They recommend a few other ones MemcacheDB, Project Voldemort, Scalaris …

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: