Re: [JAIN-SIP-INTEREST] Scalability in NIST

Fran,

The implementation uses threads - you can pre-configure a thread pool 
size or have an unlimited supply of  threads.For UDP transport, if you 
have no limit on thread pool size, a new thread is spawned per request. 
Locks are not held for significant periods of time. Other than that, the 
other source of serialization is that the NIST implementation holds a 
per transaction semaphore when the listener is executing to ensure 
integrity of the Transaction state machine. Experiments have shown 
decent scalability 1 processor achieved something like 15 caps and 4 
processors achieved 50 caps at one point of time. The performance should 
have improved since then, however, your milage may vary. There are some 
obvious optimizations I wish I had the time to attempt (such as the way 
in which messages are being converted to String, lazy parsing etc. which 
can significantly improve performance). Continue this thread on the 
nist-sip mailing list if you have an interest. This discussion list is 
only for the API and not for the NIST implementation thereof.

Ranga


 Francisco Javier Tellez Vaquero wrote:

> Hello,
> I'd like to know what about scalability with NIST. If we obtain, for
> example, 50cps with 2 cpu's, if we increase the number of CPU's to 4, 
> are we
> going to obtain near 100cps?
>
> Best Regards,
> Fran.
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Visit the JAIN website at http://java.sun.com/products/jain
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>


-- 
M. Ranganathan 

Advanced Networking Technologies Division,
National Institute of Standards and Technology (NIST),
100 Bureau Drive, Stop 8920, Gaithersburg, MD 20899. 
tel:301 975 3664 , fax:301 590 0932 http://w3.antd.nist.gov/
Advanced Networking Technologies For the People!