[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: GSE IDs [Re: IETF multihoming powder: just add IPv6 and stir]



On donderdag, mei 8, 2003, at 11:50 Europe/Amsterdam, Brian E Carpenter wrote:

and it involves stateful distribution of mapping information. A very
different beast from GSE, and it sets off my stateful=bad alarm.

Actually this wouldn't be a problem at all since we have to keep this
exact same state anyway in order to map the other way around for
sending packets back.

Again, not in GSE as I understand it.
I don't think it's a coincidence that there hasn't been any progress with GSE for five years or so. In theory, GSE can work without a mapping mechanism, but this opens the door to security problems. So in practice we need to keep state to know whether there is a valid locator <-> identifier mapping to avoid trivial identity theft. And if we accept that, we may as well remove the whole globally unique lower 64 bit thing as it just breaks too much stuff without any real benefits at this point.

Aside from that, not having a mapping mechanism makes failover very difficult: the only way that still works is if the border router at the source sees the problem. This works for last mile problems, but not for routing problems further upstream. I know others have different experiences, but for me routing problems are the number one cause of outages.

Is there anyone who wants to stick with GSE without a mapping mechanism?