This is a follow up to the last post in which we discussed performance improvements on our XPath functionality obtained from the revamped XPath and XSLT support in Mule 3.6. This time, we’ll go through the same exercise but for the XSLT use case.
Just like with XPath, we worked with Luciano Gandini from the performance team.
Sometimes (more often than we think), less concurrency is actually more. Not too long ago, I found myself in a conversation in which we were discussing non-blocking architectures, tuning, and performance. We were discussing that tuning for those models often starts with “2 threads per core” (2TPC). The discussion made me curious about how Mule’s batch module would perform if tested by 2TPC. I knew beforehand that 2TPC wouldn’t be so impressive on batch, mainly because it doesn’t use a non-blocking threading model.
Today I’m going to share some valuables lessons learned about developing highly concurrent software. These are real life lessons that come straight from the development of the Mule ESB. This is a story about deadlocks, context switches, CPU usage and profiling, focusing in how to diagnose this issues which is often the hardest part of the solution.
So the story begins a couple of weeks ago when I started working on a new feature for Mule 3.5.0.