[c-nsp] strange RTT increase in ASR1006

[Tassos Chatzithomaoglou]

Nick Hilliard wrote on 3/2/2013 01:07:
> On 02/02/2013 11:33, Tassos Chatzithomaoglou wrote:
>> The same happens in the following scenario on another ASR1006 (RP2/ESP20/SIP10); when the
>> output of Te1/1/0 goes above 6G, RTT for packets flowing from Te1/0/0to router's loopback
>> increases by 50-100ms (RP is at ~30% all the time).
>>
>>                    +-----+
>> Te1/0/0 (6G/1G) ---|     |---Te1/1/0 (1G/6G)
>>                    |     |
>>                    |     |
>>                    +-----+
> Whatever about the problems which are occurring with the esp40 + the 3x10G
> interfaces, an ESP20 should easily be able to handle 7G (=6G+1G) of simple
> in/out traffic without sweating.  Microbursts cannot explain this scenario
> because you cannot burst higher than 10G line rate ingress on either side.
>  So if there is buffering followed by packet loss and if the box is issuing
> internal flow control signals, that suggests something very peculiar is
> going on.
>
> Have you come across this document:
>
>> http://www.cisco.com/en/US/products/ps9343/products_tech_note09186a0080af2d14.shtml
> It should show you where the packets are being dropped (e.g. spa, sip or
> esp).
That was my very first source of information. Although it's missing some explanations,
it's nice to see such info on CCO.
The problem is that increased rtt (probably due to excessive buffering/queuing) isn't
discussed into this, only packet drops.

--
Tassos