[Simgrid-user] Ordering of flows

Kurt Vanmechelen kurt.vanmechelen at ua.ac.be
Tue Jul 1 11:51:24 CEST 2008


Hi,

First of all thanks for the prompt feedback on GTNetS and the action delays.
We will keep the model in which we call back to SURF on each reported delay.
I will retry the GTNetS compilation this week and let you know if everything 
works fine.

I have a new question though. As our tests show, there is no flow "ordering" 
in SURF. Suppose we have two hosts A and B,  and A starts two communication 
flows X and Y to B. Flow X is started at time U and flow Y at time V with U < 
V. Now it appears that Y can finish before X, i.e. there is no (packet/flow) 
ordering to ensure that X finishes before Y. 

If I am not mistaken the TCP protocol prevents such reorderings from happening 
by using sequence numbers. I also know that SURF is not supposed to do packet 
level simulation. 

I was wondering however whether you came across this issue before and how you 
deal with this in message-based programs (i.e. programs that used your MPI 
layer for example). In such programs it is important for this ordering to be 
correct in order to assure a semantically sound execution of an RPC based 
program.

It is certainly an option for us to attach sequence numbers to our flows and 
do the correct reordering ourselves (although this will require some more 
resources).

Thanks a lot in advance for the info.

Best regards,

Kurt

-- 
================================================
Kurt Vanmechelen
Phd Student
Dept. of Mathematics and Computer Sciences
Group CoMP (Computational Modelling and Programming)
University of Antwerp
Middelheimlaan 1
2020 Antwerpen, Belgium
G2.14

Phone : 03/265 3477
Mail     : kurt.vanmechelen at ua.ac.be
================================================




More information about the Simgrid-user mailing list