Friday, October 26, 2012

Akiban, new database, supported by SQLAlchemy

By Vasudev Ram

Akiban is a new type of database, with what may be some interesting features.

Seen via the Planet Python Twitter account, specifically, via this tweet:

It's the author of SQLAlchemy, Michael Bayer, talking about starting to provide SQLAlchemy support for Akiban.

The Akiban team seems accomplished. Akiban, the company, is based in Boston.

P.S. I was interested to see that one of the Akiban team members, Jack Orenstein, was a founder/architect at Archivas, which was later acquired by Hitachi Data Systems. I had read about Archivas some years ago; they created a kind of archival system for large amounts of data, with fast retrieval times. They used Python as one of their development tools. Archivas was also based in Boston.

P.P.S. Jack Orenstein of Akiban (and Archivas), referred to above, made a comment on this post. Check it out below. Pretty interesting info ...

Incidentally, I learnt from his comment that Jack is also the creator of Osh, the Object Shell, which was one of the tools I blogged about in my first post on some ways of doing UNIX-style pipes in Python. And seeing those tools is what inspired me to later create pipe_controller. Talk about serendipity ... :-)

- Vasudev Ram - Dancing Bison Enterprises


Jack said...

Hi Vasudev, this is Jack Orenstein. I built the "metadata manager" at Archivas. That component is a Java distribution layer over a Postgres instance on each node. It is what would now be called a NoSQL system. I became a big fan of Python at Archivas, developing a tool for doing distributed command-line and scripting operations on clusters of nodes. (It's available at The system we developed has thrived at Hitachi, and is now sold as the Hitachi Content Platform.

I was fortunate to meet the founder of Akiban, Ori Herrnstadt, as he was putting the company together. The basic idea, a hierarchical storage organization with an ordinary SQL interface was very attractive; an incredibly simple idea that somehow, everyone else had missed. We're also doing novel work in the area of indexing (e.g. multi-table "group" indexes, hybrid spatial/non-spatial indexes), and document access to SQL databases. Download it and try it out.

Vasudev Ram said...

Hi Jack,

Thanks for commenting. Your info is quite interesting.

Cool to know that you were also the creator of osh. I had blogged about it (along with other such tools) here, a while ago:

And seeing those tools is what inspired me to later create pipe_controller:

which I blogged about here:

1) PipeController v0.1 released

2) Using PipeController to run a pipe incrementally

3) Swapping pipe components at run time with pipe_controller

Thanks, I'll check out your products.