GroupStudy.com GroupStudy.com - A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
Re: bandwidth question [7:49002] posted 07/18/2002
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]


Dear priscilla

Thanks for your reply :)

Well....the 10% packet loss happen at the peak time...and that can happen
for a period of 3-4 hours

No packet loss was observed during off peak hours.

I rememeber reading something on a cisco article which states that WAN
performance will worsen when it gets over 70 % utilised. As for my link it
is around 96-98% utilised so i guess the cisco guideline is right ...



""Priscilla Oppenheimer""  wrote in message
news:200207171714.RAA14182@xxxxxxxxxxxxxxxxx
> birdy wrote:
> >
> > Can anyone tell me why....
> >
> > I have 2 Mbps WAN connection that reached 95 % utilisation
> > during peak time.
>
> For how long was it at 95%? That would definitely worry me, unless it
turned
> out that "the peak" was just a few seconds or something.
>
> > When I try to ping to my provider serial interface(next hop),it
> > register 10%
> > packet loss
> >
> > My router serial interface is showing 1.9Mbps (incoming
> > traffic). Since the
> > router is receiving 1.9 Mbps out of the 2M pipe,which means
> > that there is
> > still a balance of 0.1 Mbps (100 k) and so therotically
> > speaking, there
> > should not be any packet loss.
>
> Pings may not be a good way to measure packet loss because the provider
may
> rate limit its responses to pings. But even if that's not the case, there
> could be packet loss when the average utilization was 95%. Over how much
> time was the average calculated? There could have been times when the
> utilization was 50% and there could have been times (like right when you
> were sending the pings) when the utilization was 100% and packets had to
get
> dropped. That could still work out to an average of 95%.
>
> >
> > I think is time for me to upgrade my bandwidth....
>
> That could be true, but you may want to do a more detailed study over a
> longer timeframe, (unless users are already making a stink).
>
> Good luck.
>
> Priscilla




Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=49078&t=49002
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to abuse@xxxxxxxxxxxxxx