Ray, I reviewed every post that has been posted on the NANOG list for the past two days, and nada concerning an outage in the region which you cited. But that doesn't mean that one didn't occur. The reporting on cable cuts and lost routes is not mandatory where the ISPs are concerned. Actually, the only obligations that they have, along with all other service providers including the LECs and IXCs, in reporting outages are keyed to POTS metrics.
In other words, the bogie is something like 30,000 lines if I'm not mistaken. That is, if 30,000 POTS accounts (or loss of subscriber sessions, as would occur in event of a SS7 signaling outage) that are knocked out of service. Those are POTS considerations, mind you, not Internet accounts.
When an outage affects both the required number of POTS lines, AND also affects ISP accounts, then it is a reportable event. But not when it affects ISP affairs, only. Priorities? These were originally spelled out about ten years ago by the National Reliability Council, whose name has been changed in recent years to something else, which slips my mind at the moment. Anyone? I think it's ATIS.
[[Late edit: The new name is, in fact, ATIS, standing for Alliance for Telecommunications Industry Solutions. I need to refresh on some of the criteria which I referred in this message. You can do so too, by researching atis.org ]]
Basically, the NRC, or ATIS, if that's what they are now called, is a cooperative made up of leading industry players (mostly the Top IXCs and ILECs) who report to Congress and the FCC I believe, and was formed as a result of a series of catastrophic failures resulting from the first fiber cuts that occurred on LD systems, and the flurry of SS7 failures which occurred during late Eighties and Early Nineties.
When you think about it, the number of ISP accounts --or stated another way, the number of Internet users whose sessions have been affected-- could not be measured very easily in any consistent way, anyway, at least not based on a severed cable or a router failure, or a problem in a dwdm device, say.
It is especially problematical to measure the real time, or even near recent term effects of cable cuts on Internet sessions, because ISP router algorithms usually make allowances for routinely restarting sessions, and routing around congested spots and bogged down resources, anyway.
"The general reason I'm curious is that I would have thought that this link would have had a redundant backup to cover. We had none."
It would seem that way, and may very well have been as you say. But you don't know that for a fact. The problem that existed might have been a logical one. As such, if the same logic was programmed into both the primary and secondary set of route provisions, as often occurs, then the problem being common to both would have kept you down just as long as if you didn't have a backup at all. You really don't want to know how many times this actually occurs. This is not an uncommon event, in other words.
If I hear anything I'll let you know.
FAC |
