Re: A tunneling proposal

[Iljitsch van Beijnum <iljitsch@muada.com>]

On Mon, 16 Jul 2001, Daniel Senie wrote:

> >One can think of a simple extension to tunneling under such ISP-wide
> >outages---if tunnel creation fails, the second ISP initiates
> >non-aggregatable route announcements for the prefixes from the address
> >space of the first ISP who has become unreachable.

Speaking from operational experience: this is not going to work. As a
multihomed site, I have to regularly check if I can still reach the entire
internet when one line goes down, because it wouldn't be the first time
someone placed the wrong filters at the wrong place. If you don't test
something like this, it's not going to work when you need it.

Also, when huge amounts of more specific prefixes appear, this could be just
the kind of thing to drive a lot of routers over the edge, especially when
we're getting close to a common limit, such as 128 MB in Cisco routers.

And when you can announce more specifics some of the time, why not make
"some" "a lot" or even "all"? And then people will start to filter and we're
back to square one.

> >For example, I know that, in the context of server farms, about 20% of
> >outages and errors are because of local power failures, but this may not
> >hold in the ISP case. Does anyone have any info about the main problems
> >and their frequencies?

> Where's that 20% figure come from? With all the backups the colo vendor 
> salesfolk talk about, that's a surprising figure.

Power is getting less and less stable these days. Those anti-globalisation
people do have a point, you know.

I got this from http://www.apcc.com/power/problems.cfm

"Power problems are the largest cause of data loss"

Power Failure/Surge: 45.3% 
Storm Damage: 9.4% 
Fire or Explosion: 8.2% 
Hardware/Software Error: 8.2% 
Flood & Water Damage: 6.7% 
Earthquake: 5.5% 
Network Outage: 4.5% 
Human Error/Sabotage: 3.2% 
HVAC Failure: 2.3% 
Other 6.7% 

Source: Contingency Planning 

So power problems are about 8 times as common as earthquakes. That's not so
bad.

Iljitsch van Beijnum