It's rare to find an IT organization that does not have a business-critical system or application that isn't quite as nimble and lightfooted as the developers and users imagined it to be.  And that's an understatement. So what do you do if you have a sluggish application? Call the database designer  to tune the database? Or assume it has to do with the application server and  call in the vendor to tune the settings? Or perhaps hand it off to the network administrator to check the network latencies? Or do all of this since you have some money left over from your outsourcing budget that your clever offshore vendor saved you?

If you have a system or application that is not performing to expectations, you need to find out where the bottlenecks are before you call someone to tune the system. Sounds somewhat obvious, but that's what is not usually done, because doing that requires a cross-platform view - something that is not straightforward and requires a certain level of  specialized quantitative analysis. PEA employs these specialized analysis techniques to isolate bottlenecks at the system and application level, and tells you whether it is hardware or software related.  If it is software related, we tell you whether the fault is with the application code (and which part of it) or a software platform. Once the bottlenecks are isolated, it is a relatively easy matter to call in the appropriate platform/product experts to tune the bottleneck away, or hand over to the development team to fix the issue.

If yours is a high-traffic Web-based application, you'd be surprised to know that the biggest army of tuners you can assemble will often turn out to be inexplicably ineffective - and that's because the problem may be with the peculiar bursty nature of Web traffic itself, and not with the system at all. Whether this is the case is something we concretely establish before getting into system performance analysis.

Would you believe our approach actually costs less than the "tune it like mad" approach?