Mailing List Archive: Ping RTD behaviour

Ping RTD behaviour

Nov 20, 2002, 1:38 AM

Post #1 of 3 (1505 views)

Hi all

Can someone help understanding the following ping RTD behaviour, the RTD is
always better beween R1 and the SERVER than between R1 and R2 (directly
connected)

R1
-------------------------------R2--------------------Switch-----------------
----SERVER
10.0.0.1 10.0.0.2
20.0.0.2

When I perform a ping test between R1 ip@10.0.0.1 to R2 ip@10.0.0.2 the
RTD is :round-trip min/avg/max/stddev = 0.818/0.865/0.968/0.043 ms

When I perform a ping test between R1 ip@10.0.0.1 and the SERVER
Ip@20.0.0.2 the RTD is always better , round-trip min/avg/max/stddev =
0.533/0.578/0.679/0.052 ms

Regards
Abdel

Ping RTD behaviour [ In reply to ]

gtate at juniper

Nov 20, 2002, 10:23 AM

Post #2 of 3 (1437 views)

Permalink

When you ping the router R2 you are sending packets to the Routing
Engine. R2 has queues from the forwarding plane (PFE) to the control
plane (RE). Then the reply needs to be processed by the RE and then
sent back to the PFE (Queued again). The link between the RE and the
PFE is FastEther. This is where the extra delay is probably introduced

The server/switch is probably a faster path than the PFE-RE one.

Gary

Abdelhamid TIBERDANI wrote:
> Hi all
>
> Can someone help understanding the following ping RTD behaviour, the RTD is
> always better beween R1 and the SERVER than between R1 and R2 (directly
> connected)
>
> R1
> -------------------------------R2--------------------Switch-----------------
> ----SERVER
> 10.0.0.1 10.0.0.2
> 20.0.0.2
>
> When I perform a ping test between R1 ip@10.0.0.1 to R2 ip@10.0.0.2 the
> RTD is :round-trip min/avg/max/stddev = 0.818/0.865/0.968/0.043 ms
>
> When I perform a ping test between R1 ip@10.0.0.1 and the SERVER
> Ip@20.0.0.2 the RTD is always better , round-trip min/avg/max/stddev =
> 0.533/0.578/0.679/0.052 ms
>
> Regards
> Abdel
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp@puck.nether.net
> http://puck.nether.net/mailman/listinfo/juniper-nsp

Ping RTD behaviour [ In reply to ]

gtate at juniper

Nov 20, 2002, 12:06 PM

Post #3 of 3 (1437 views)

Permalink

Just to add to this - echo requests are given the lowest priority so
that any routing updates packets, management packets etc are processed
first and are placed in a higher priority Q than echo replies.

Gary

Gary Tate wrote:
> When you ping the router R2 you are sending packets to the Routing
> Engine. R2 has queues from the forwarding plane (PFE) to the control
> plane (RE). Then the reply needs to be processed by the RE and then
> sent back to the PFE (Queued again). The link between the RE and the
> PFE is FastEther. This is where the extra delay is probably introduced
>
> The server/switch is probably a faster path than the PFE-RE one.
>
> Gary
>
> Abdelhamid TIBERDANI wrote:
>
>> Hi all
>>
>> Can someone help understanding the following ping RTD behaviour, the
>> RTD is
>> always better beween R1 and the SERVER than between R1 and R2 (directly
>> connected)
>>
>> R1
>> -------------------------------R2--------------------Switch-----------------
>> ----SERVER
>> 10.0.0.1 10.0.0.2
>> 20.0.0.2
>>
>> When I perform a ping test between R1 ip@10.0.0.1 to R2 ip@10.0.0.2 the
>> RTD is :round-trip min/avg/max/stddev = 0.818/0.865/0.968/0.043 ms
>>
>> When I perform a ping test between R1 ip@10.0.0.1 and the SERVER
>> Ip@20.0.0.2 the RTD is always better , round-trip min/avg/max/stddev =
>> 0.533/0.578/0.679/0.052 ms
>>
>> Regards
>> Abdel
>>
>> _______________________________________________
>> juniper-nsp mailing list juniper-nsp@puck.nether.net
>> http://puck.nether.net/mailman/listinfo/juniper-nsp
>
>