kernelnewbies.kernelnewbies.org archive mirror
 help / color / mirror / Atom feed
* Measuring Network Packet processing time
@ 2014-04-07 13:21 Sunny
  2014-04-08 13:35 ` Peter Senna Tschudin
  0 siblings, 1 reply; 3+ messages in thread
From: Sunny @ 2014-04-07 13:21 UTC (permalink / raw)
  To: kernelnewbies

Hi all,

I?m running xen 4.2 on Linux Kernel 3.7 and I?m trying to figure out the time taken(latency) by a network packet to reach it?s destination VM from the time it arrives at the NIC. What do you think is the best way to do it? I was thinking may be add rdtsc timestamp to the IP options header on arrival, and one more when it reaches to its desired VM and examine the two timestamps. I?ve not figured out a way to do this yet?but do you think this is a viable option or what other methods would you recommend?

Thanks,
Sunny

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Measuring Network Packet processing time
  2014-04-07 13:21 Measuring Network Packet processing time Sunny
@ 2014-04-08 13:35 ` Peter Senna Tschudin
  2014-04-08 15:33   ` Valdis.Kletnieks at vt.edu
  0 siblings, 1 reply; 3+ messages in thread
From: Peter Senna Tschudin @ 2014-04-08 13:35 UTC (permalink / raw)
  To: kernelnewbies

You need to know exactly what you are doing for using rdtsc. There are
some external factors that affect the counting such as dynamic clock,
and out of order execution. I'm not sure that rdtsc registers are
synchronized on different CPUs. Some code for you to play with:
https://github.com/petersenna/rdtscbench
https://github.com/petersenna/Kernel/tree/master/stopwatch

On Mon, Apr 7, 2014 at 3:21 PM, Sunny <sundarcs@gwu.edu> wrote:
> Hi all,
>
> I'm running xen 4.2 on Linux Kernel 3.7 and I'm trying to figure out the time taken(latency) by a network packet to reach it's destination VM from the time it arrives at the NIC. What do you think is the best way to do it? I was thinking may be add rdtsc timestamp to the IP options header on arrival, and one more when it reaches to its desired VM and examine the two timestamps. I've not figured out a way to do this yet--but do you think this is a viable option or what other methods would you recommend?
>
> Thanks,
> Sunny
> _______________________________________________
> Kernelnewbies mailing list
> Kernelnewbies at kernelnewbies.org
> http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies



-- 
Peter

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Measuring Network Packet processing time
  2014-04-08 13:35 ` Peter Senna Tschudin
@ 2014-04-08 15:33   ` Valdis.Kletnieks at vt.edu
  0 siblings, 0 replies; 3+ messages in thread
From: Valdis.Kletnieks at vt.edu @ 2014-04-08 15:33 UTC (permalink / raw)
  To: kernelnewbies

On Tue, 08 Apr 2014 15:35:29 +0200, Peter Senna Tschudin said:
> I'm not sure that rdtsc registers are synchronized on different CPUs.

There exist Intel multi-core chipsets that explicitly do *not* have
synchronised rdtsc registers.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 848 bytes
Desc: not available
Url : http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20140408/bf4905ae/attachment.bin 

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2014-04-08 15:33 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-04-07 13:21 Measuring Network Packet processing time Sunny
2014-04-08 13:35 ` Peter Senna Tschudin
2014-04-08 15:33   ` Valdis.Kletnieks at vt.edu

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).