Hiranya has written an excellent blog post on How the World's Fastest ESB was Made. This post clearly articulates the making of the pass-through transport.
The post also clearly describes the limitations that were there in the old NIO transport, where there were two shared buffers used and also the use of Axiom and Synapse between the two buffers.
Then it explains how the new pass-through transport was designed ground up to overcome the limitations and bottlenecks of the old NIO transport, using a one shared buffer, and how it by-passed the Axiom layer.
If you read the blog by Hiranya, you will get a very clear picture not only about the high level design, but also the various challenges faced and how they were overcome.
The latest ESB performance round is a solemn testimony of what this design is capable of.
Comments