Re: Finetuning the TX-ring values posted 03/06/2006
If you were trying to add something to the technical discusion, I'm afraid I
missed the point of your post :)
On 3/5/06, Scott Morris <swm@xxxxxxxxxx> wrote:
> That's a very nice way of saying "There are programmers that are paid to
> know a lot more of the finer nuances of what the heck a TX-Ring is and
> it does to make sure that most people do NOT need to mess with them!".
> -----Original Message-----
> From: nobody@xxxxxxxxxxxxxx [mailto:nobody@xxxxxxxxxxxxxx] On Behalf Of
> Chris Lewis
> Sent: Sunday, March 05, 2006 10:56 PM
> To: Amit Dass
> Cc: ccielab@xxxxxxxxxxxxxx
> Subject: Re: Finetuning the TX-ring values
> IMHO it is not useful to look at the tx-ring in isolation. When a network
> designed for a given end to end latency, the fixed and variable latencies
> should be within the goal, be it 150 msec for voice or whatever the
> is designed for.
> IOS self tuines the tx-ring size to 2 for links less than 2 Mbits, 4 for
> links less than 10 Meg, 8 for links less than 20 Meg and so on. It is only
> worth tuning the tx-ring to less than the default, when the default does
> meet the end to end design goal (and don't tune to less than 2). It is
> practice to leave the tx-ring at the highest value that will still meet
> delay requirements as that helps the interface to maintain line rate
> transmission under a broader range of offered load conditions.
> On 3/5/06, Amit Dass <amitdass2003@xxxxxxxxx> wrote:
> > Hi Group,
> > How much is the Tx-rings finetuning helpful in avoiding the
> > excessive serialization delay.
> > What are the optimum values of Tx-rings on a T1/E1 link.
> > Thanks in Advance.
> > Best regards,
> > Amit
> > ---------------------------------
> > Yahoo! Mail
> > Bring photos to life! New PhotoMail makes sharing a breeze.
> > ______________________________________________________________________
> > _ Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
> Subscription information may be found at: