[ih] Network "cutover" techniques - 1984 Old School

[Louis Mamakos]

I recall when I was at UUNET, and we were acquired by a success of
phone companies.. I found myself in a meeting with the optical transport
people trying to plan out future needs.  This was in the mid to late
1990's during the period of Internet hyper growth.   How much do
we need?  How much do you have?!

I described how often we added capacity in our backbone trunking, which
at the time was in the OC3, OC12 days, adding capacity every few weeks
as circuits got delivered.  I saw looks of surprise, and got asked
"When do you schedule all the down time?"  Huh?  Downtime?  "Yeah, when
you add the links, don't you have to update all the route tables
in the switches?"  Uh, no, the routers just sorta figures that out all
on their own, dynamically.  Like how it recovers from link failures.
"No, really, when do you do that?"

They really couldn't believe all the connectivity was that dynamic, 
which
was more surprising than how many more T-1's of capacity we needed.
Their gut instincts from years and years of voice telephony-based
thinking was not serving well, for the times, they were a changing.

louie


On 7 Apr 2021, at 15:32, Dave Crocker via Internet-history wrote:

> On 4/7/2021 12:21 PM, Jack Haverty via Internet-history wrote:
>> It took just "a few seconds" to complete the transition - a major
>> improvement to the typical times of "minutes or even hours" that was 
>> the
>> norm.
>
> In 1985, I worked about a large IBM SNA operation.  Having only the 
> Arpanet as framework for doing networking, I was not prepared to hear 
> that addition or removal of a node required taking down the entire 
> network...
>
> d/
>
> -- 
> Dave Crocker
> Brandenburg InternetWorking
> bbiw.net
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history