job: architecture review of a scalable erlang system
Thu Oct 1 22:34:32 CEST 2009
I wrote a scalable Erlang system for my client. The system can
handle hundreds of thousands of clients on EC2 and uses up to
200 instances in simulation.
The system supports both socket and http connections and has
certain latency requirements which are fully met by the socket
side. The HTTP side of the system does not give me the latency
required by my customer but we are in a disagreement over whether
the architecture of the HTTP transport is causing this.
I spent a couple of months tweaking the system and I don't think
I can push HTTP performance any further. I need an expert
review of the architecture and suggestions for performance
improvement. The system is ~7K LOC including an extensive
simulation harness and is written by me entirely in Erlang.
This is a short-term paid job and only experts need apply.
Please contact Neil Berkman at with a description
of highly scalable Erlang code that you have written and optimized.
Please copy me on your email message.
fastest mac firefox!
More information about the erlang-questions