🚀 go-pugleaf

James E Robinson (jerobins@unity.ncsu.edu) wrote:
: 
: I do believe you fail to recall the numerous problems with the MCI link
: to Atlanta.  So bad that MCI moved the link from Atlanta to
: Greensboro....at least our link to DC via Sprint has been reasonably
: stable as compared to MCI, if only we could get a packet *though* DC,
: now that would be cool.

That core1 router in G'Boro that MCI runs has got to be the flakiest 
piece of equipment on the face of the planet. As to the MAE-EAST problem 
in DC, the peering point up there is at capacity and getting ready to 
melt down. For significant stretches of the late afternoon and evening 
eastern time they are running at 100 percent of capacity.  That is why 
one needs to go with a level one provider who has private peering 
arrangements as well as those at the NAPs. That way there is always a 
route-around.

One thing I find amusing with Sprint is how convoluted their routing 
actually is between any two given points. I have some customers who are 
hung off of a level three provider on a dial-up who call me when they 
have problems getting to their web site (as if it is my fault.) The first 
thing I do is a trace route - especially if they are on a Sprint or 
Alter.net network and look at the path. There is not one time I ever 
remember seeing less than eight routers between my client and the NAP. 
And half of them seem to be flapping most of the time.  :)

Steve Crisp
PagePlop Web Hosting Service
http://www.pageplop.com