Customer AS

Perry E. Metzger perry at piermont.com
Tue Aug 20 16:22:54 UTC 1996


"Justin W. Newton" writes:
> >1) redundancy in all components and links.
> >2) No "cowboy"ism.
> 
> 3) These sites aren't turning up /hundreds/ of T-1's and other links on a
> given week.  

Ahem. How does that change the fact that router changes are frequently
done "cowboy style"? How does that change the fact that one frequently
calls a NOC and gets the comment "sorry, but *THE* router in Philly is
down".

> 4) Points of entry into these networks are very limited, and therefore the
> network manager has much more control over what comes into and out of his
> network as far as both data and routing advertisements.

Again, this neither addresses the cowboy issue nor the lack of redundancy.

> 5) These networks are not nearly large enough at this point in time to
> seriously push hardware to/beyond its design specifications.

Again, this neither addresses the cowboy issue nor the lack of
redundancy.  The majority of the time my clients have outages caused
by non-redundant hardware failures or cowboyism at providers, not by
hardware being pushed too far.

> 6) Traffic patterns are relativelty predictable, you don't suddenly have
> triple the traffic over one link that you didn't have yesterday.

Again, this neither addresses the cowboy issue nor the lack of
redundancy.

I'm sure that all of what you mention are indeed problems -- problems
that will lead to missed install dates, slow networks, etc. None of
those, however, are the problems that I'm talking about.

Perry





More information about the NANOG mailing list