Allow me to fill in (see below).
On Tue, 15 Feb 2000, Ron Buchalski wrote:
>>From: Hank Nussbacher <hank@att.net.il>
>>To: cisco-nsp@puck.nether.net
>>Subject: PA-T3 question
>>Date: Tue, 15 Feb 2000 07:10:49 +0200
>>MIME-Version: 1.0
>>
>>Running 12.0(5)S on a pair of 7505 routers with a PA-T3 in each.The PA-T3
>>card is:
>>
>>TAU-gp2#sho contr serial
>>Serial0/0/0 -
>> Mx T3(1) HW Revision 0x3, FW Revision 2.56
>>
>>The physical line is a satellite T3 line.Occassionally, the RTT starts to
>>jump from the normal 560ms and reaches 1400ms.It starts its climb over a
>>period of times and can may take a few hours of slow climbing till it
>>reaches unbearable levels and someone notices it and does a reset.Only a
>>'clear interface' will make the line act normal again.
>>
>>The physical line and modems have all been checked and the suspicion now
>>lies on IOS or the PA-T3 cards.
>
>Are you plugging directly into a satellite modem, or are you using a
>terrestrial (telco) circuit to get to the satellite earth station?It
>almost sounds like you have a clocking problem on the satellite link.
There are local loops at both sides.
>Even though the satellite is geosynchronous (fixed position), it's actual
>distance to the earth varies throughout the day.Because of doppler
>effects, the data rate changes as well.Normally a point-to-point satellite
>link will be set up with one end clocking the circuit as the master, and the
>other end set to loop the receive clock (slave) and use it to transmit back
>to the master site.On the receive side at the master site should be a
>receive buffer that will absorb the rate variations throughout the day.
>
>If the buffering is missing, or screwed up, it may manifest itself in what
>you're seeing.The PA-T3 does not have adequate (physical layer) buffering
>to handle a satellite link.This should be handled by the satellite service
>provider.
The modems do have anti-Doppler-effect buffers. In fact, the link wouldn't
work without them for more than a few minutes. Also, the magnitude of the
RTT deviation (a second, and sometimes even more) is much too large with
respect to Doppler effect in this environment (buffers of several mili-secs
worth of traffic are enough).
So the question remains: where do the packets spend this extra second? Note
that if packets are not lost (I suspect at least some do) than at T3 rates
a second is quite a bit of buffer space.
If you know of a site with a similar configuration, please let us know, and
we'll check with them to see what they experience.
Thanks,
Oded.
>
>-rb
>
>______________________________________________________
>Get Your Private, Free Email at http://www.hotmail.com
>
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:10 EDT