Server overload does not come from network overload, but from service overload. If you can't serve (compute) more requests/s than the network load, your server is overloaded... You can compute your server overload by queuing requests. If your queue becomes too big, you can consider that your server is under heavy pressure and take appropriate measures to circumvent the flow of requests (load balancing etc...) 2011/1/4 Avi Bahra

I am experimenting with boost asio & serialisation. and have created a test server based on the corresponding example in the asio tutorial.

This is a simple single threaded server. The stress testing I have done, shows I get different results on different platforms. This is expected.

However I need a way of determining when the server can not cope with the load. This allows me to inform the user.

Currently I have used typical requests to stress test the server. This shows the server handles ~1000 requests/sec in debug under ideal conditions, on localhost.

Does this means that if I continue to receive 1000/request sec over a several sec/minutes, then the server must be overloaded. Is this the right way of thinking about the problem or am I "barking up the wrong tree".

Can any one provide any insights about how one goes about determining when a asio server is overloaded ?

Any help appreciated.

Best regards, Ta, Avi _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users