EXPERT RESPONSE
Achieving high availability in DB2 (or with any DBMS) is a constant challenge. At its very core, availability can be broken down into four distinct components all of which combine to assure that systems are running and business can be conducted:
1. Manageability -- the ability to create and maintain an effective environment that delivers service to users.
2. Recoverability -- the ability to re-establish service in the event of an error or component failure.
3. Reliability -- the ability to deliver service to specified levels for a stated period of time.
4. Serviceability -- the ability to effectively determine the existence of problems, diagnosis their cause(s) and repair the problem.
All four of these "abilities" impact the overall availability of a system, database or application.
To assure manageability, there are two basic rules:
1. Be sure to design your databases and applications with availability in mind.
2. Make sure you have the right tools for the job.
Designing for availability means that you take maintenance into account as you build your databases. Consider things like partitioning, the number of NPIs and the impact of these on reorganization. If you design properly, you can maintain partition by partition while leaving other partitions available.
Also, having the right tools is a BIG consideration. Make sure that your reorganization utility can function while data remains online and accessible by your applications. Further, make sure that the REORG can function (to completion) with the amount of concurrent activity you will have in your apps. Some REORGs do not scale well if concurrent modifications are being made while the REORG runs. Additionally, a change management solution is imperative in a high availability situation. Without a change manager, the amount of time it takes to make changes to your databases increases and so will downtime.
Additionally, make sure that you have the proper recovery management solution so that you can quickly react to problems and recover DB2 data rapidly when problems occur. Things to look for include transaction recovery capabilities (undo/redo SQL), backout capability, recovery simulation and automation.
Good luck.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our discussion forums.
|
