Canonical Database Architecture and DB2 Performance … Really?

by Susan Lawson, Daniel L. Luksetich
May 13, 2009

Getting the Best of Both Worlds

A canonical database architecture can work; it can give the enterprise a common methodology for accessing various disparate data stores and processes using messaging and a common communication pattern. It gives the enterprise the ability to quickly build interfaces and react to changes in business.

Is this easy to do? Maybe. Is it the best for performance? No way! You can, however, achieve both commonality and performance if exceptions to the enterprise rules can be made for high-performance situations. Most applications can be written or retrofitted to the enterprise architecture without high-performance costs if they’re generally low-volume applications that don’t move much data. Also, for most mission-critical, high-volume applications, you also can use the enterprise layered design. However, for the high-volume portions of the application, a custom design can be used to deliver huge performance gains. Applying these exceptions to the rule can achieve the desired performance and this approach typically involves a small amount of code that’s accessed most of the time. It gives you all the benefits of enterprise architecture without the major performance implications.

As always, there are trade-offs to consider, and you must remain aware of the specific needs of your applications and database to succeed in these environments.