RE: QOS Two Rates

From: Andy (unk.student@gmail.com)
Date: Sun Mar 23 2008 - 09:13:46 ART


Hello!

After reading some posts from GS archives and RFCs about policing I suppose
that policing do not use concept of Tc.
Number of tokens (bytes), which allowed to drain from Bc bucket defined by
interpacket delay.
Following formula is used:
(Current_Packet_Arrival_time - Previous_Packet_arival_time)*CIR/8.*

* W. Odom CCIE R&S Exam Certification guide, Third edition, p 490.

By default Bc bucket size calculated through the following formula:

Bc = (CIR * 0.25 sec) / 8 bit/byte = CIR/32.
One exception exist Bc < 1500 bytes, then IOS set Bc = 1500.

There is great post from Bruce Caslow in GS archives about policing.
 http://www.groupstudy.com/archives/ccielab/200611/msg01631.html.

But I still confused with the tasks, where Bc value given in bps, not in
bytes. What is the formula for such calculation?

-----Original Message-----
From: Sadiq Yakasai [mailto:sadiqtanko@gmail.com]
Sent: Sunday, March 23, 2008 12:54 AM
To: Andy
Subject: Re: QOS Two Rates

Hi Andy,

You see, CIR is normally in bps, but Bc is normally in bits or bytes.
Bc values are not rates,

Hence the formular: CIR = Bc/Tc, where Bc is in bytes (or bit) and Tc
is in seconds (or milliseconds).

Sometimes, you might be asked to regard Bc value in bps. This clearly
relates to what your Tc value would be.

Lets say you are asked to regard Bc as 8000bps, CIR as 32000bps. You
need to know what Tc is, or at least know how many Tc intervals you
have in every one second. Then you can get how many Bc bits you send
in every Tc interval and from this, you would know how many Tc
intervals you would have in 1second.

I am actually on my way out clubbing with some friends and they are
kinda yelling at me but hope it helps.

If it isnt clear enough, just ping me back and we can hypothesize some
more on it.

Sadiq



This archive was generated by hypermail 2.1.4 : Tue Apr 01 2008 - 07:53:54 ART