Nice article.

However, there’s one DB you missed mentioning: ArangoDB.
I’d like to share my little story with it and also present some facts surrounding it.

In search for a database that would fit my needs and having evaluated some of the ones mentioned in this article I found a very promising new kid on the block.
Don’t get me wrong. I have used MongoDB in the past and evaluated OrientDB for a short time. Both are very good Dbs each with their great features, but didn’t fit my needs for the project I have been working on.
While ArangoDB doesn’t yet meet one primary need of mine, which is transaction support, it will provide it very soon, in Version 1.2 according to the roadmap. You’ll find out below, why I chose it, nevertheless.

Besides the great features and amazing ideas that are implemented in this database, I came to really appreciate the openness and forthcoming of the developers on any questions I threw at them during my evaluation. The experience with the ArangoDB developers came to a high point when I proposed some enhancements and features. The devs checked if the features had meaning and could be done without introducing problems to the db. They implemented the suggested features mostly on the same day! That’s something I had never seen before with any other DB. Having had that nice experience I just had to give back to the project and so I started contributing, by helping out, with the ArangoDB PHP API.

I must say, it’s been a very nice experience so far and it has all the signs that it will continue to do so.

So, finally, here are some facts on ArangoDB:

ArangoDB is an open-source, multi-threaded most flexible all-in-one database solution.

You can mix “key-value store”, “document store” and “graph database”, all in the same DB.

Indexes: You have “Full-text”, “Geo”, “Hash”, “Bitmap” and “Skip-list” indexes to choose from. You have the freedom to use what you need for your specific application.

Querying: There are several ways to do queries in ArangoDB. Simple queries by id or example and ArangoDB’s Query Language AQL for the more complex queries, also supporting PATH queries on Graphs.

ArangoDB also features Actions, which allow you to write server side code in Javascript and provide a REST interface to call them.

It makes use of shapes to store and reuse common document schemas instead of writing same document attributes in every document over and over again. This results in far less space usage in memory and disk, which automatically translates in increased performance.

Speed: Here’s a nice one! ArangoDB uses HTTP REST as its interface. So, one would think that would be slow…. This Blog post (http://www.arangodb.org/2012/10/04/gain-factor-of-5-using-batch-updates) will prove quiet the opposite. The tests that were run in that post show that up to 500.000 document inserts in less than 5 seconds are possible. This depends of course on the hardware, document size and various other factors, but bottom line: It’s amazingly fast!

MVCC: ArangoDB uses AppendOnly/MVCC to update documents. That way locks are minimized and documents are quickly appended to the datafiles without having to reorganize or find empty spots, etc… Of course there is garbage collection which gets rid of the replaced documents.

Drivers: Since access to the DB is through the REST interface there are no binary drivers. Actually one could write his/her own driver if it doesn’t yet exist. Up-to-date there are drivers for PHP, Node.js, Ruby, Java, Perl, Python & D.

ArangoDB 1.1 will soon be released with many new features and performance improvements.

Transactions will be introduced in Version 1.2. Replication and sharding are on the roadmap for Version 2.0.

So, in the end ArangoDB is definitely a database to keep an eye on