On Fri, 25 May 2001, Robert Davies wrote:
> Quite probably, problems I saw were poor and unreliable routing, regular
> unreachability with UUnet customers (sometimes this was UUnet's fault but it
> took 3-4 days for them to track it down)
My other half pointed out that she couldn't read her webmail gadget from
work. We eventually tracked it down to a routing loop inside UUNet's
network between Reading and Bristol, and her employers' WAN terminates in
UUNet's Reading POP... which wasn't useful at all. Was nice to have it
found, mind :)
As for routing, we get our BGP table from their Telehouse router nowadays
as the one at Daleside was a little under-specced to be doing that job. I
suspect we weren't the only ones - now that the global route table has
gone over 100k networks for the first time there's a lot of older routers
complaining about RAM shortages. And Cisco RAM ain't cheap, either.
[Thinks: to get back on-topic, I wonder what difference I'd have if I
routed everything through a pair of Linux boxes running the LVS+heartbeat
patched kernel, plus gated & zebra? It'd be easier to do firewalling,
anyway...]
> I know we weren't the only customers affected, perhaps your multiple peering
> protected you from the worst, or you might have more direct routing as a
> large customer, but as a business customer it was not much fun relying on
> them.
Our multiple peering is sometimes a big PITA. If we continue to receive
route tables *but* the peer network is dead somewhere, then all traffic
going that way just goes into a black hole. Which causes us massive
problems and has now happened twice with NTL and twice with UUNet. Ah
well, that's what I get paid for!
G
--------------------------------------------------------------------
http://www.lug.org.uk http://www.linuxportal.co.uk
http://www.linuxjob.co.uk http://www.linuxshop.co.uk
--------------------------------------------------------------------
This archive was generated by hypermail 2.1.3 : Thu 22 Nov 2001 - 13:13:12 GMT