Xref: utzoo news.admin:7356 news.config:1454 Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!usc!gem.mps.ohio-state.edu!wuarchive!decwrl!atha!lyndon From: lyndon@cs.AthabascaU.CA (Lyndon Nerenberg) Newsgroups: news.admin,news.config Subject: Re: Disaster Planning Message-ID: <1195@atha.AthabascaU.CA> Date: 27 Oct 89 17:53:27 GMT References: <103@farcomp.UUCP> Followup-To: news.admin Distribution: news Organization: Athabasca University Lines: 55 In article <103@farcomp.UUCP> shibumi@farcomp.UUCP (Kenton A. Hoover) writes: >I propose that there be an disaster plan for USENET that entails its being >"turned off" in local areas during periods when utility services are >suffering excessive loading. Well, the first thing we did was shut down our link with decwrl. This is just plain old common sense. It turned out that decwrl survived and continued polling us through most of the aftermath, even though we couldn't get a line in to them. My understanding is that Pacbell simply disabled the incoming LD trunks to ensure local bandwidth would be available in the affected area. Something similar happened in Edmonton a couple of years ago when we were hit by The Tornado. In that instance, nearly all of the system administrators had the foresite to shut their modems off as well. What would you propose as a disaster plan? and how would you ensure the majority of sites were aware of it? Do you expect a "global" plan to be applicable in the majority of cases where the disaster is (relatively) local in scope? I feel that such planning is a good activity to be undertaken at the local "users group" level, since the vast majority of uucp connections are made with sites inside a local toll area. You should take into account the existance of sites with leased line connections outside the potentially affected area that might survive the disaster. These sites could become hubs capable of feeding other sites during the wee hours of the night when the load on the telephone network is substantially reduced (0100-0500). You would probably want to shut down all news except for regional groups, and perhaps have one or two people near the affected area post regular updates to the mainstream groups to try to cut down on the number of "what's going on out there" requests that inevitably flood the net. Most of this problem is due to propogation delay; As the updates flow out, they pass the sea of incoming queries. The UUCP Mapping Project was also very much on the ball during the earthquake. As sites were verified as non-operational, they were marked as dead in the d.AProject file, and updates were forwarded into comp.mail.maps on a regular basis. This allowed the well connected sites to rapidly reconfigure their routing tables. Of course, the mail had to eventually bottleneck *somewhere* (I would be interested in hearing from anyone who had this problem). Hopefully someone from the Bay area can summarize what actions were taken. This should give us a good base to start working from. I'm sure the people in Vancouver would be willing to beta test whatever we come up with in a year or two :-) -- Lyndon Nerenberg VE6BBM / Computing Services / Athabasca University {alberta,decwrl,lsuc}!atha!lyndon || lyndon@cs.AthabascaU.CA The Connector is the Notwork.