What a cool name for a project and I'd be excited to work on something like this. I'm probably not that qualified to cut any code, but the idea is pretty neat. A reporter at Information Week noticed this and the items were captured in this blog.
I haven't heard if this will be in SQL 11, but I'm sure that at some point we'll get some of this technology in the platform. It's something that many people would like to see and I can see the marketing people having a field day with the idea of "Matrix technology" embedded in SQL Server.
In thinking about this a little, I see two ways that this can be done. The first is that you automatically "partition" database resources across servers. Essentially we have the "shared nothing" model, where each server has control of it's own resources, and maintains some integrity or checks between them. It would extend what we have today with partitioned views and allow us to bring on a database that can be accessed, knowing that there's other data out there not available. However you would query the data as though everything is on one server.
That's cool, but it concerns me. How do we handle which data is where? What about lookup data? Do we have data that's on all nodes? That would certainly make some sense, and this solves one scalability issue: large data sets.
But there's another scalability issue that I'm more concerned about and I'm hoping will be tackled in SQL 11. Right now in SQL Server 2005 (and in SQL Server 2008), we have Shared Scalable Databases, which are read only instances of the database server that all access the same SAN disks with the data. Now that allows us to share the load somewhat, but without write access, it really limits the value here.
If there were one write instance and multiple reads, then it would be much better, but that doesn't seem to be available right now.
The big issue that people want to solve, at least from what I can see, is that people want to add database instances like they add web servers. Getting the instance overloaded, add another server and share the load. We really want load balanced instances and that's what I'm hoping comes with the MatrixDB project.
I know it's a hard problem to solve, and maybe it would even be one that wouldn't necessarily be the best for DBAs since you could add cheap commodity servers instead of expensive, skilled DBAs, but it's the technology I'd like to see in SQL Server some day.
Steve Jones
The Voice of the DBA Podcasts
The podcast feeds are now available at sqlservercentral.podshow.com to get better bandwidth and maybe a little more exposure :). Comments are definitely appreciated and wanted, and you can get feeds from there.
or now on iTunes!
- Windows Media Podcast - MB WMV
- iPod Video Podcast - MB MP4
- MP3 Audio Podcast - MB
Today's podcast features music by Everyday Jones. No relation, but I stumbled on to them and really like the music. Support this great duo at www.everydayjones.com.
I really appreciate and value feedback on the podcasts. Let us know what you like, don't like, or even send in ideas for the show. If you'd like to comment, post something here. The boss will be sure to read it.