Adam Rothschild wrote:
On 2008-11-02-10:14:14, Matthew Kaufman <matthew@eeph.com> wrote:
But seriously, it shouldn't be necessary to have two connections at work [...]
This is less than clear, and largely dependent on a specific organization's [in]ability to function if their internets go down. End-site multihoming in some form or fashion is a growing requirement, and folk thinking otherwise need to get their heads out of sand.
There was an implicit (and sometimes contractually explicit) social contract associated with buying from a Tier 1 provider. That was that I'd pay a premium for connecting, and that everyone could reach me. That's never a guarantee - I've had both upstream peering games and more conventional network outages in the last 20 years than I can conveniently count. But the social contract was that I was not going to be bitten by my provider playing games with other providers. If I am going to be bitten by that, then I have to take additional mitigating measures, which includes multihoming for smaller sites than I used to do. That costs money. In this particular economic climate, if I have to tell site owners that they need to pay more money in connectivity, hardware, engineering, and ondoing administration time to multihome, they are going to tell me to make it up on the back end, which means that paying for Tier 1 grade connectivity is no longer affordable. Somewhere, inside Sprint, is a salesperson who understands this and has been trying to tell his or her coworkers. Apparently to no good effect...
If anything, these recent de-peerings underscore the lack of wisdom in end users connecting to (or purchasing CDN services from) members of the tier 1 club directly.
The recent depeerings have two effects: One, they devalue higher priced Tier 1 ISP connectivity because they reintroduce the political/financial depeering risk into overall connectivity reliability, and two, it pushes the boundary downwards in both small/mid tier ISPs and content sites where multihoming is either necessary or strongly advised. That leads to additional cost and engineering time, a need to shift high level industry effort more towards supporting new multihoming sites again (ugh, this is never a fun surge to deal with, as everyone finds out the things BGP is bad for all over again). All on top of the looming IPv6 deployment surge we're all facing. Not to be alarmist, but what the @#$#@%@#(*&$% ? -george william herbert gherbert@retro.com