Intelligent Enterprise

There's more to/about the Infobright open-source announcement than I covered in my Intelligent Enterprise article. I have thoughts to share on Infobright's architecture and limitations of the release. There's more to say about the MySQL data-warehousing context and then there's the puzzle of the significantly delayed MySQL 5.1 general availability (GA) release.

Infobright is a MySQL storage engine. MySQL runs on many computing platforms. The first bit left out of my article is that the software released today runs only on 64-bit Linux. According to Infobright VP of Marketing and Product Management Susan Davis:

"We expect to release ICE [Infobright Community Edition] on 32-bit Linux within the next 30 days. Given the memory limitations of a 32-bit server however, we recommend the use of a 64-bit server for production use. We believe that the 32-bit version will make it easy for people to download and try it however, so that is our motivation for releasing a 32-bit version. We have no plans to release IEE [Infobright Enterprise Edition] on 32-bit due to the memory constraints and the resulting performance hit. At this point we do not have a definite timeframe for Windows, but will be guided by our customers (for IEE) and community (for ICE).
"We are planning to port Infobright Enterprise Edition to Solaris. As far as porting ICE to OpenSolaris, we will see what input we get from the community in terms of importance."

I've been using MySQL for over a decade, and I'd love to try Infobright myself, but my server runs Windows 2003. I'll have to wait. And the response regarding OpenSolaris was interesting. Sun Microsystems owns MySQL and just invested millions in Infobright. Sun CEO Jonathan Schwartz says "OpenSolaris scales on any hardware, has built in virtualization, great web service infrastructure, fault management, diagnosability, and tons more." But for Sun, perhaps recognizing marketplace reality, it still takes a back seat.

I've also been looking forward to the general availability release of MySQL 5.1, which Sun announced as "pending" back last April. I asked MySQL spokesperson Steve Curry's about the delay. Here's his response, verbatim:

"The answer is that we've considered 'shipping' the product twice in the last six months but decided after internal discussion that it would be better to fix some more bugs. Although these remaining bugs would not impact the vast majority of MySQL users, and we could've released solutions in subsequent software updates, we decided to hold off and fix them now instead.
"The MySQL engineering team is still a rather small one by industry standards, and with the acquisition by Sun, we hope to have access to additional resources and experience that will allow us to improve and streamline our future product development process.
"I think it's important to point out that this situation is a little different than a traditional proprietary product. Since MySQL is an open source product under an open development model, anyone can download the current beta software and decide for themselves if it is ready for use. Even though we do not recomend 'pre-GA' software for production systems, we have ourselves been running MySQL 5.1 on our own Web site (not a small one) for over a year.

Release 5.0 went GA in October 2005. Three years is a long wait for refresh of software as widely relied on as MySQL. I wrote a Pentaho-sponsored white paper on 5.0 "MySQL: Ready For Primetime Business Intelligence," back in early 2006, but 5.0 simply doesn't match rival DBMSes as a DW platform. I've been eager to see GA release of 5.1 features including partitioning and indexing and performance improvements that should boost the MyISAM engine's data-warehousing suitability. (Do note that Infobright's storage engine is an alternative to MyISAM.) Looking further down the road, according to spokeperson Curry, "MySQL 6.0 is slated for sometime next year."

Then there's Kickfire MySQL appliance, which would seem to compete with Infobright. Kickfire's "patented SQL chip... uses parallel, pipelined data flow to pack the power of tens of high-end general purpose CPUs into each chip." The appliance supports MySQL data warehousing that relies on the standard MyISAM storage engine and also accelerates use of the InnoDB transaction oriented storage engine. There wasn't room to mention Kickfire in my article, nor compelling cause really: Kickfire targets databasing with up to 3 terabytes of user data whereas Infobright claims to scale up to 30-50 terabytes.

Lastly, regarding Infobright's technical underpinnings — data packs and the knowledge grid semantic layer — IE editor Doug Henschen and I agreed to leave the following puzzlement of mine on the cutting room floor: "It is unclear how data packs and the knowledge grid significantly differ from the statistical optimizer, operating on storage pages, first deployed commercially by the Ingres RDBMS in the last 1980s. A possible interpretation is that Infobright has applied decades-old query-optimization techniques to its compressed column-store DBMS." Old ideas can be good ideas.

So there you have it: the rest of the Infobright, Kickfire, MySQL 5.1, and MySQL platform story.

E-MAIL | FOLLOW US