gutierre@oblio.arc.nasa.gov (Robert Gutierrez) (02/21/90)
djcl@contact.uucp (woody) writes: > A while ago, I called an 800 number that told me that the number could > not be called, but suggested I try 1 800 888.1800 for assistance. > (This was dialed from 416 area code, and I think it worked from 519 as > well). > Well, I would try 1 800 888.1800 only to get the recording (from MCI > presumably) that the call could not be completed. The recording then > suggested for assistance that 1 800 888.1800 be dialed for further > assistance. You got timed out while the local MCI switch was polling the 2 VAX's they use for 800 number lookup. What happens is that the switches have dedicated data lines to 2 VAX's, one in West Orange, NJ and the other in Richardson, TX. The switch has a 'primary' VAX to poll, and if there is no response, then the other VAX is polled. What has happened in the past is that one VAX has crashed, and the other VAX has to take up slack, but the response time is so slow that the switch polling it just times out the call and routes it to that recording. And then there was this one time when both VAX's died for 45 minutes.... Oh yes, 800-888-1800 routes to Customer Service most of the time. Robert Gutierrez - NASA Science Internet Network Operations.
c186aj@cory.berkeley.edu (Steve Forrette) (02/22/90)
In article <4144@accuvax.nwu.edu> you write: >You got timed out while the local MCI switch was polling the 2 VAX's >they use for 800 number lookup. >And then there was this one time when both VAX's died for 45 minutes.... If it is true that just 2 VAXes (or the leased lines going to them) going down can bring down the entire MCI 800 system, perhaps AT&T's promise of 1 hour service restoration for downed 800 lines isn't all that half-baked...
rrw@cci632.uucp (Richard R. Wessman) (02/23/90)
In article <4144@accuvax.nwu.edu> Robert Gutierrez <gutierre@oblio.arc. nasa.gov writes: >What happens is that the switches have dedicated data lines to 2 >VAX's, one in West Orange, NJ and the other in Richardson, TX. The >switch has a 'primary' VAX to poll, and if there is no response, then >the other VAX is polled. What has happened in the past is that one VAX >has crashed, and the other VAX has to take up slack, but the response >time is so slow that the switch polling it just times out the call and >routes it to that recording. A minor point (if you don't work for CCI), the machines at the DAP's in New Jersey and Texas are not VAXes. They are POWER 6/32 FT's, which are fault-tolerant versions of POWER 6/32's. For the lawyers: POWER 6/32 and POWER 6/32FT are registered trademarks of Computer Consoles, Incorporated. Rick Wessman cci632!rrw
gutierre@nsipo.arc.nasa.gov (Robert Gutierrez) (02/26/90)
c186aj@cory.berkeley.edu (Steve Forrette) writes: > >You got timed out while the local MCI switch was polling the 2 VAX's > >they use for 800 number lookup. > >And then there was this one time when both VAX's died for 45 minutes.... > If it is true that just 2 VAXes (or the leased lines going to them) going > down can bring down the entire MCI 800 system, perhaps AT&T's promise of > 1 hour service restoration for downed 800 lines isn't all that half-baked... In another article, Rick Wessman correctly points out that these computers are CCI Power 6/32-FT's. These are fault-tolerant computers, and as such, they are less prone to 'go down'. But, as was proved in AT&T's fiasco, just because the computer has double back-ups, that doesn't mean the code loaded into them can't go bad. If I remember correctly, that was the problem in MCI's 45 minute 'outage'. MCI, almost one year ago, upgraded the DAP's ("Data Access Point") to do number-table updates every 15 minutes instead of every night, so that emergency rerouting could be offered to their customers. If a customer is having a local problem, and has a pre-defined emergency plan submitted, he or she can call MCI and give a password, and can have that emergency plan implemented within 15 minutes (the next update to the DAP's). The software to do this was apparently very tricky, since updates and lookups were being done almost the same time now. Also, the software upgrade was to do call-restriction down to the NPA-NXX level, as opposed to just the State level before that. Also, it was the beta test of calling-number delivery. As with the switches, the hardware may be excellent, but the software sometimes can have problems....especially if you have 1E+32 patches to it..... Robert Gutierrez NASA Science Internet Network Operations Moffett Feild, California. "You know, all of my patients had one thing in common...... ....They all hated the phone company." James Coburn, "The Presidents Analyst".