A challenge of end-to-end application response time measurement is that different client applications demand such different tuning strategies. For example, I was recently working with a company whose web tier was composed entirely of Java VMs in Linux. Measuring the end response time involved using web testing software, virtualization management software, and more. The strategy we used there is completely different than the client I was at this weekend, who did analytics with a thick client front end.

It sounds like you’re ready to move up past the free events and graduate to more senior ones like pre-conference and post-conference events that really dive deeply into a subject instead of doing a lot of one-hour wading.

I’ve been to many SQL Saturdays, viewed PASS videos, and read a lot of books. I’d obviously learn more in some focused training events. But, I have seen many excellent presentations. In Grant Fritchey’s PASS presentation he matter-of-fact answered one question I’ve had and asked, but never got a clear answer until his.

I prefer presentations that focus on one specific strategy, e.g. user response time, managing server bottlenecks, etc. I think K.D. covers the general strategies in her 2k5 performance tuning book.

I have yet to see a good presentation on end-to-end application response time tuning, where we proportionally measure and show where the user’s business task waits from SQL Server perspective. I’m talking beyond SQL statement tuning, e.g. to measure the waits events like network latency, cpu, io, locks, blocks, etc. I’m interested in wherever the time is going, proportionally.

Also, I’d like to see more presentations of index evaluation and comparison used by query groups, rather than looking for one SQL statement’s missing index. Lets assume and discuss a list of queries and evaluate the set of indexes they used. Make sense?

HTH