Mon, 13 Feb 1995 21:29:52 +0100
|
On Mon, 13 Feb 1995 11:18:30 CST Patrick Douglas Crispen
<[log in to unmask]> said:
>Are these assumptions correct?
>
> 1. If UGA gets flooded with INTERBIT traffic, UGA's mainframes will
> start to slow down from the load. If UGA's mainframes slow,
> Bitnet mail routed through UGA will also slow because the
> mainframes will be preoccupied with INTERBIT traffic.
True.
> 2. If Bitnet mail through UGA slows, Bitnet mail net-wide will
> slow.
>
> 3. If UGA crashes from the load, Bitnet will crash too.
Well, all mail through UGA would slow. Most importantly, after a certain
amount of impact on computing resources which were not originally
purchased to run BITNET services, it becomes difficult for management to
justify continuing to provide the service. UIC for instance decided to
stop providing INTERBIT services. When this happens, the files have to go
somewhere else. It was very hard to find a "somewhere else" for UIC and I
don't think there can be a "somewhere else" ever again, because all the
core machines are overloaded now.
> 4. Bitnet users subscribed to LISTSERV lists at non-Bitnet sites
> (e.g. Texas A&M) will see a noticeable delay in mail delivery to
> and from the list.
Currently yes, but this will stop being an issue with the 1.8b release.
It will only affect a subset of the subscribers, too.
> 5. Internet users who used the percent hack to subscribe to a
> LISTSERV list at a Bitnet site will also experience delays.
Possibly.
> 6. A big majority of the INTERBIT traffic is actually from sites
> such as CompuServe and AOL who subscribe to Bitnet list
> addresses.
If you meant commercial providers, I don't think so. They account for a
lot of traffic, but probably not a "big majority".
> 7. The natural consequence of all of this is that if the big net
> trafficers moved up to Listproc or LISTSERV TCP/IP then the
> INTERBIT problem would be severely reduced.
In general if people stop using INTERBIT, INTERBIT will stop being a
problem. Some large sites are already starting to do that as a result of
the message I sent and maybe the storm will be averted.
> 8. If the big users are forced off of the Bitnet, they will abandon
> the Bitnet -- and CREN -- altogether.
Well, that's up to them. It's a standard cost/benefit decision.
Eric
|
|
|