Re: BGP/4 (fwd)

S. Van Steen (steven@glo.be)
Sun, 8 Dec 1996 00:27:03 +0100 (MET)

> > True. But you need to be able to switch and reroute packets at a fast
> >speed. And I really don't think that a 386 with 4 megs of ram will hold a full
> >BGP table.
>
> Well, the IRX and PM-2 can go to 16 Megs right now. If you can find 30 pin
> SIMMS larger than 4 Megs each then I guess we might be able to have the code
> address the larger size.

Currently having a full BGP path and routing table on a cisco 7500 in dual
homed config will eat nearly 32MB making 64MB an absolute must have. In any
case current portmaster hardware is not capabable of dealing with such
enormious routing tables let alone simply using 2 fully loaded T1 lines.
IRX might one day become usefull at peering points (would also be a dream to
see IS-IS or EIGRP by then) but backbone routing no way, never.

Taking in account the time it took between promising VLSM/OSPF and actualy
seeing it (still waiting for pm2er30) my guess is that Livingston will have
BGP4 running by the time it is ancient history, I'd rather see some effort
in ipv6 and the mysterious stuck-W1 bug (deny everything you want but it
remains a comos bug)

--Steven.