[c-nsp] ipsla - latency - related to cellular backhaul

Chuck Church chuckchurch at gmail.com
Fri Apr 26 10:41:59 EDT 2013


We use a decent software based router dedicated to IP SLA to source all
probes from (3845 in our case).  If you have a device at the destination
that has consistently low CPU (like an L2 switch that doesn't get clobbered
with SNMP constantly), that works good for ICMP probing.  For UDP jitter
type testing, you need an IP SLA responder.  If it's that important, you can
put a dedicated router out there (871 are dirt cheap on EBay).  

Chuck

-----Original Message-----
From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Pete
Lumbis
Sent: Friday, April 26, 2013 9:56 AM
To: Tony
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] ipsla - latency - related to cellular backhaul

Some hardware platforms and offload ping, mainly Echo Reply. I know that
ASR1k and the GSR can do this off the top of my head (that is, I'm not
saying this is an exhaustive list). Echo Requests will always be generated
by the Supervisor/RP/Central CPU.

If Echo Replies are not offloaded then the possibility of jitter/higher
latency always exists and will not be a valid test compared to data plane.
If a data plane will be handled through hardware, while we have to punt a
ping to the CPU then schedule the IP Input and ICMP processes (in IOS) to
handle these. The time it takes to schedule this will very if there are
higher priority processes running. If you sit at your desk and ping your
local gateway router you will probably see a fair amount of jitter. If you
ping something else that isn't a router, or that can offload ICMP, you'll
see both better response times and less jitter.

-Pete


On Thu, Apr 25, 2013 at 7:35 PM, Tony <td_miles at yahoo.com> wrote:

> Hi,
>
>
>
>
> >________________________________
> > From: Aaron <aaron1 at gvtc.com>
> >
> >Tac says that this drop and the latency seen using various ipsla 
> >pings is expected since all pings are treated less than everything 
> >else and could
> be
> >getting policed by LPTS (I don't know what LPTS is)
> >
>
> Google tells me that LPTS = Local Packet Transport Services. TAC are 
> meaning packets that are destined for the router control plane, not 
> the forwarding plane (ie. packets TO the router, not THROUGH the router).
> Response to these packets can depend on how busy the router is and 
> also any CoPP that might be implemented. Has potentially to be true. 
> If you have no CoPP on the devices and they are under minimal load 
> (CPU wise) then this probably shouldn't be a factor.
>
> Are you losing any traffic that is going through the device (ie. from 
> ping
> tests) ?
>
>
> regards,
> Tony.
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net 
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/



More information about the cisco-nsp mailing list