[erlang-questions] CPU load of TCP server

Andrey Tsirulev <>
Wed Oct 14 14:15:53 CEST 2009


Hello Max,

Well, I cannot agree syntetic tests are so useless as they can show at least 
theoretical limit for my conditions (before I even started to implement real 
business logic).

I'm not sure I understood the opposition of erlang virtual machine to erlang 
environment, could you explain it, please? Now I prepare a release and start 
erlang with a boot script, what is another approach?

Andrey

----- Original Message ----- 
From: "Max Lapshin" <>
To: "Andrey Tsirulev" <>
Cc: "Rapsey" <>; <>
Sent: Wednesday, October 14, 2009 12:03 AM
Subject: Re: [erlang-questions] CPU load of TCP server


> Andrew, You should test against real business logic, because syntetic
> tests don't show anything at all.
>
> While experimenting with video streaming server, I've got strange
> results: 5 users used the same CPU, like 30.
> So I advice to test not erlang virtual machine (which isn't fastest
> one), but erlang environment (which can
> be faster due to advanced binary usage and scalability possibilities). 



More information about the erlang-questions mailing list