Well, many many people have been bombarding me with the new paradigms that they claim will leave the traditional SQL engines in the dust. Hard to believe that, with a 20+B industry, but nonetheless, every week brings a new one... To name a few, Google's BigTable (internal use), CouchDB, and now, SimpleDB by Amazon. All promising to simplify persistence, with a simple query model. I think all are interesting developments, but comparisons to SQL databases are meaningless here.
I like them all, but for different reasons. BigTable, of course, as an abstraction for map-reduce class of applications. CouchDB, which counts on many fans, including IBM's own Pat Mueller and Sam Ruby (and I am becoming one!), written in Erlang, is very elegant in its capability/simplicity tradeoff. And now, Amazon, perhaps not satisfied by the availability of structured persistence through eco-systems of the like of Gigaspaces, decides to offer SimpleDB, which is not quite Dynamo (at least in my reading), and is also in Erlang (any relationship to CouchDB? -- I will have to do some more reading), for applications in the cloud.
Watch this space for our moves in this arena...
Heya Arant,
SimpleDB and CouchDB are not related. It is currently believed that SimpleDB uses Amazon's Erlang interface to BDB.
I wholeheartedly agree with the statement that comparing all these projects to relational databases is meaningless. They are another solution for a similar set of problems, but surely not the same and there are things where SQL beats anything else hands down.
Also, welcome to the CouchDB-fan-zone :)
Cheers,
Jan
--
Posted by: Jan | December 17, 2007 at 01:01 AM
Anant --
Just to clarify, GigaSpaces plays a very different role than something like a SimpleDB does. GigaSpaces is an application platform -- a distributed application server, if you will -- that lets you built a stateful application that can easily scale across a distributed environment such as Amazon EC2. GigaSpaces manages your business logic, messaging and data. For data it does so through an in-memory data grid, and provides a Persistence as a Service feature that would persist the data to a persistent store such as SimpleDB. So the two solutions are complementary.
Regards,
Geva Perry
GigaSpaces
Posted by: Geva Perry | December 17, 2007 at 05:10 AM
To second Geva i would suggest looking the following blog posts on my blog:
Putting the Database Where It Belongs
PaaS – Persistence as a Service (using Hibernate)
Posted by: Nati Shalom | December 17, 2007 at 11:18 AM