Common causes of performance problems and outages

Tue Jul 3 12:04:13 UTC 2001

At 01:27 03/07/01, Roeland Meyer wrote:
>In no particular order;
>
>1:Peering polcies resulting in really bizarre business-dictated routes.
>2:Over-booking on broadband links (cable-modems)
>3:Over extended growth by DSL operators (w/ resultant business failures, ie
>NorthPoint).
>4:RBOCs ... ALL of them.
>5:Fixing what ain't broke.
>6:Forgetting why you did something and then doing it "right" later.
>7:ICANN for broken DNS (which BIZ you iz).
>8:NewNet, for REALLY proving 7:.
>9:The A**holes that write scripts for script-kiddies.
>10: ...I'll stop while I'm behind....

s/2:/2a:/
s/on broadband links/on HFC segments/
2b:     DSL operators wildly oversubscribing the link 
        between the DSLAM and the first-hop router.

2a/2b are equivalent problems, but the location of 
oversubscription risk varies with the technology.

11:     Deployment of "advanced" technologies that a reasonable
        NOC cannot make work in one's backbone/core.