High Availability Guide for DB2

Having seen Chris talk at several IDUG events, I knew he had a deep knowledge of DB2 and its high availability features. This book highlights that knowledge. The section on logging is the clearest explanation of DB2 logging that I have seen anywhere. I found it especially useful that the parameters that affect logging are covered in detail. His chapter, "Minimizing Unplanned Outages" is full of good information about increasing data availability during required maintenance. His detail regarding online schema changes is quite helpful. The chapter on setting up standby database gives much good detail about the various techniques. Sample scripts are included in the chapter to make the DBAs good easier. The chapter on HACMP is very helpful for a DBA trying to understand how HACMP works. My only disappointment with the book is that it came out too early to cover HADR. Having a section on HADR would have made the book complete.

If you have been a sysadmin and are familiar with the concepts of full and incremental backups of an operating system, then this book's descriptions of dB2's abilities in this field are very impressive. dB2 offers far more fine grained control/resolution over the backing up of its data. Like the means of having circular or archival logging. So if you are just starting out as a dB2 administrator, you have quite a bit more to learn, just in this field.On a tangential note: Suppose you have your own non-database application, with its own native data file format, and you want to regularly save its activities. Try looking at the ideas in this book for how dB2 does its duties. You might be able to abstract out the ideas and reinstantiate them in your context. The book presents the right level of discussion for this. One caveat is that IBM has a portfolio of patents around dB2, and I do not know the extent that you might be able to implement some of the book's ideas without infringing on the patents.Of course, there is far more to dB2 than backups. The book goes into other topics. Like how to define and run a standby database, by sending log files to it in a systematic fashion. Or, how to make a "split mirror" warm standby. My overall impression is that IBM has engineered a vast amount of backup and availability into dB2, to handle almost any level of transaction load. There is a lot more to dB2 than just being able to handle standard SQL.