Closed powerman closed 12 years ago
I think this may be related to different load level on AWS or network issues between AWS servers, but numbers in https://github.com/ericmoritz/wsdemo/blob/results-v1/results.md looks just wrong.
I've run 30 second tests on my workstation. In theory, my results should be less correct because both client and server was running on same system. But results looks much more like something I expect to see:
handshake latency messages connections timeouts Erlang 4.008 3.595 203259 10000 0 Go 2.291 1.289 204493 10000 0 Perl 2.266 1.937 204526 10000 0
Another strange thing - I expect these tests will load all 8 cores, but actual load was only about 60% (i.e. ~5 cores).
I think you are probably correct. I am rewriting the event collection mechanism so that the raw data is available for detecting flaws.
I think this may be related to different load level on AWS or network issues between AWS servers, but numbers in https://github.com/ericmoritz/wsdemo/blob/results-v1/results.md looks just wrong.
I've run 30 second tests on my workstation. In theory, my results should be less correct because both client and server was running on same system. But results looks much more like something I expect to see:
Another strange thing - I expect these tests will load all 8 cores, but actual load was only about 60% (i.e. ~5 cores).