Wednesday, January 30, 2008

Postgres in the Sky (well, EnterpriseDB in the Cloud)

EnterpriseDB Corp announced yesterday that they will be joining Amazon in the cloud. So what the heck does that mean?

Amazon's cloud computing is also called the Amazon Elastic Cloud Compute or Amazon EC2. EC2, when tied with Amazon's cloud storage (Amazon S3), gives you cheap and easy scalability for your applications. Your computing power moves from your own data center and hardware to the "cloud". You control the servers but they are virtual servers running somewhere else. You control access, you control what applications are running and who can use them. If you only need a trickle, you pay for a trickle. When you need to scale to huge proportions, you pay for what you need.

Amazon recently announced Amazon SimpleDB which utilizes Amazon EC2 and Amazon S3. With SimpleDB, you can store and query data (i.e. build data web services). I really wasn't all that impressed with SimpleDB when I first read about it as it is a very simple database. It's named well. ;-)

SimpleDB gives you named value sets of data. It is the most basic of data structures. You can assign anything to the "database" by giving it a name and a value. A name can have multiple values. And so on.

EnterpriseDB's announcement, EnterpriseDB to Deliver OLTP Database Using Amazon Cloud, means that instead of the very basic SimpleDB, you will now be able to get a robust, Enterprise-class database in the cloud.

The beta testing for EnterpriseDB Advanced Server Cloud Edition starts in March, 2008. Cool.

You can run EnterpriseDB on a tiny server to build and test your apps and then, in minutes, scale up to huge proportions when you go live. No worries about data centers or buying hardware.

No mention if the vanilla EnterpriseDB PostgreSQL package would be offered. I would think not as EnterpriseDB Advanced Server is more geared towards this kind of solution.

I'm not sure what kind of SLAs Amazon offers. It will be interesting to see what kind of guarantees can be offered on a service like this. OLTP processing requires very reliable computing, much more so than OLAP and reporting.

We live in interesting times.

LewisC