[AWS] Having more then 1000 concurrent connections?

[Zhu Qun-Ying]

On Fri, 26 Aug 2011 03:48:36 -0700, Marius Amado-Alves  
<marius at amado-alves.info> wrote:

>> Software engineering does not exist in a vacuum, it relies on hard  
>> matter
>> (silicon, usually) that sits below it.
>
> No. Most program designs rely only on the language. Yes, the language
> is a model of the machine, normally via operating system.
>
>> The problem is - the traffic capacity is at least in some degree
>> *determined* by the number of cores (why do you think companies buy  
>> bigger
>> servers to handle bigger traffic?),
>
> indeed one common method of improving the performance of a system is
> just to throw more hardware at it (instead of reprogramming)


I think this is too ideal a condition.  You will constant facing some  
limitation impose by OS or hardware resources and adding new hardware is  
not always the solution.

And if it is targeting embedded system, memory usage is a big concern  
also. Lots of tasks mean lots of memory usage and at some point there will  
be no memory left for other processes in the system.

>
>> which [actually validates, not] blows off your "abstraction layers"  
>> argument.
>
> We agree that
>
> Maximum_Number_Of_Tasks = Expected_Traffic * Number_Of_Cores *  
> Some_Constant
> _______________________________________________
> AWS mailing list
> AWS at lists.adacore.com
> /no-more-mailman.html