[Spread-users] Bad performance small packets

M.B.M. Leoné m.b.m.leone at student.utwente.nl
Fri Jan 5 06:20:43 EST 2007

Hello everyone,
For a research project I'm comparing the performance of Spread and other
related systems. In this test I measure the delay between sending a request
to a host and getting a response back from this host.
The request is sent with multicast, the response gets back unicast. I'm
using the Java API under Windows XP and use 2 hosts for this test.
When using Spread in this test I get a very strange result, small packets of
512 and 1024 bytes get a much higher delay then larger ones.
The packets of 512 and 1024 bytes have a (standard) total delay of about 195
ms, larger ones have only a delay between 8 and 19 ms.
This very large difference cannot be caused by my own implementation, with
other multicast systems this problem doesn't occur and I use the standard
Spread operations.
It seems that it is caused by the way Spread handles small packets. Is this
standard behaviour of Spread or is it perhaps caused by a bug?
I hope some of you can give an answer on this question.
Best regards,
Marco Leoné
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.spread.org/pipermail/spread-users/attachments/20070105/fd104299/attachment.html 

More information about the Spread-users mailing list