I have used BMC Apptune the most, but also IBM CQM and CA SQL
Detector (a few years back). All have some good features. All
probably have weaknesses.
I think you need to work out what features of BMC Apptune that
you rely on heavily and are important to your site, and then check
if the competition products can do the same things. Above all the
price per year is going to matter a lot I would think, and level of
support you get when having problems.
Also look at the flexibility of the product in terms of being
easily able to turn on or off certain expensive sampling features
for a short time when you need them, e.g. Data Object Analysis of
Object Level Getpages and Waits.
Does the product allow you to cut back your sampling during peak
hours, to a bare minimum or even turn off sampling for a known peak
Review the overheads of the product, not so much the cost in the
address spaces used by the monitor, but the effect on Package CPU
consumption, caused indirectly by the product sampling. You need to
capture Db2 Acctng data from SMF and monitor it historically to see
the effects of turning on and off the monitoring product. Can be
very high sampling costs for very frequent SQL calls that do only
very small work. Ability to tailor the sampling is important. No
use to have a product that costs to much in CPU making it
impractical to use. Important to be aware of the real indirect
overheads. Even Db2 Class 2 and 7 (without one of these monitors)
can be a significant overhead on some workloads. Db2 does allow
these to be turned off selected areas, with well understood
For BMC Apptune I have seen a paper reporting the approximate
sampling overhead costs on a per SQL and Per Getpage basis with
different sampling options. Would be great if someone actually had
sampling overhead figures for all the competitive products. The
products can change at a new release where the vendor tries to
decrease the sample overheads despite the ever growing increase in
measurement fields possible at each new release of Db2.
I have not done a direct competition on all 3 products to really
see which one has the least overheads inside the Packages nor to
compare all the features so I don't know which product is best.
The important features for me off the top of my head (but can be
different for each site):
1. Getpages used on an individual SQL call basis. Also I/O waits
and various types of lock or contention waits especially for Data
2. Ability to capture Host Variable Values on exceptions, extra
high CPU or Getpages SQLs.
3. Ability to turn on Object Analysis only when needed.
4. Ability to concentrate Dynamic SQLs captured by pattern
(ignoring the variation in literal constants)
5. "What if" feature to quickly test if an index change or an
SQL change could greatly improve the access path (and estimated
costs - taken with a grain of salt).
6. Breakdown of CPU into zIIP and General processor CPU.
Breakdown of Dynamic SQL into Prepare CPU and execution
7. Above all a good price of course!
DB2 Application Performance Specialist
CPT Global Ltd