Robert Cannon <rmcannon@mail.com> writes:
Focus Group III of NRIC is the Focus Group working on this issue. Subcommittee 2 is looking at future considerations. This is the subcommittee that came up with the recommendation to have a trial period of ISP outage reporting.
Of course, some large companies will continue their current practice of keeping bad news quiet. But my intent hasn't been to point out how bad some companies are. My concern is the wild, unregulated Internet gets blamed for a lot of outages which really aren't "Internet" outages. In the last two weeks I've only noticed three multi-hour multi-ISP affecting incidents: 1) a misconfigured router in Dallas spewed bogus routing information, 2) an ILEC's DACS lost its configuration in New Jersey, and 3) a MAE had a loopback which caused "network congestion" and routes to flap for a few hours. There were other problems, a couple of fiber cuts and power failures, but they didn't seem to cause problems for multiple ISPs. Interestingly enough problem 1 affected multi-homed networks more than single homed networks, which happen also to house the most popular web services. With so much hype this month about web retailers and the christmas shopping season, some very high profile sites have been hit by various problems and covered by the news media. However, just because a credit card processor goes down, and many web sites were shutdown, does not mean its an "Internet" problem. Its a credit card processor problem. As I poked around several of the problems over the last couple of weeks, I found most web site outages related to "back-end services." Problems with their database software, disk and server farms, and the ever popular "operator error." Yes, some network problems happened. But ISPs are getting pretty good at the packet delivery stuff. The ISP can usually deliver more packets than the server can handle :-) Perhaps surprising, but since some folks are concerned e-commerce is where the money is, none of the problem e-commerce web sites with whom I spoke or read about attributed any down time to criminal or "cracker" activity. But it may be a self-selecting group. Maybe the reason some people wouldn't answer my questions was because their outages did involve suspicious activity.