RE: [nsp] tx-ring-limit tuning on ATM PVC

From: Kinczli Zoltán (Zoltan.Kinczli@Synergon.hu)
Date: Wed Nov 21 2001 - 03:21:46 EST


Hello,

  I'm not aware of any rule, but if you adjust the bandwith parameter
correctly
the tx-ring-limit seems to reflect that adjustment.

  In my experience on low speed links/PVCs tx-ring-limit of 2 or 3 was
OK.

  Just an idea: think about the delay/jitter as a function of
tx-ring-limit: the higher
the ring-limit is, the later the QoS takes place. If you have a ring
limit of 10, than
10 packets of max MTU could be 'sitting' in the hw FIFO without
triggering the QoS meachanism.
Given the bit rate of the line/PVC it gives you a time.

  Someone from Cisco might correct me if i'm wrong...

regards,
  --Zoltan

-----Eredeti üzenet-----
Feladó: Francois Baligant [mailto:francois@be.wanadoo.com]
Küldve: 2001. november 8. 3:44
Címzett: cisco-nsp@puck.nether.net
Tárgy: [nsp] tx-ring-limit tuning on ATM PVC

        As anyone a mathematical method to tune the tx-ring-limit
        parameter on an ATM PVC ?

        Right now Im using a packet generator and checking if Im
        dropping packets with 'sh queuing int atmx/x.x' when the
        PVC is slighty full.

        I would like to keep the tx-ring-limit as low as possible
        to be able to do low latency queuing for some particular
        QoS class.

        I must pass all my traffic throught one PVC on that ATM
        sub-interface. I have applied a policy-map to the PVC
        to choose what I want to drop first of course.

        Francois

-- 

Francois Baligant _ Wanadoo Belgium NV/SA, Network Operation Center ( ) a subsidiary of France Telecom /_\/ Lozenberg 22 - B-1932 Zaventem francois@be.wanadoo.com (__/\ tel: +32 2 717 17 17 FB1-6BONE FB3122-RIPE fax: +32 2 717 17 77

- "if you hold a unix shell to your ear, do you hear the c?"



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:24 EDT