NAP/ISP Saturation WAS: Re: Exchanges that matter...
Chris A. Icide
chris at nap.net
Fri Dec 20 17:15:51 UTC 1996
>> And you don't find the 7500 a big jump over the 7000?
>> What are you trying to do, IP tunnel your data through another backbone?
>Ok, yes they added some memory and it is faster, but they needed to put a
>MUCH bigger CPU in that box. It was a jump, but not as big as the one we
>all needed. The 7500 was not any new technology, just adding more of the
>current. And no we are not tunneling through another backbone, we run our
>backbone over Cascade 9000 with clear channel DS3s.
Definitely, the 75XX is a jump over the 70XX router platform. However, it
is very simple to collapse a 7513. The RSP2 does not have enough
processing capability to allow economical use of the 11 available slots.
We've been able to collapse several 7513's with less than half of the
slots filled and heavy tuning by Cisco Engineers.
Now, I am in no way, form, or manner an authoritative Cisco information
source, however, I have heard some rumors which I place fairly strong
1. Cisco will be releasing a new RSP card soon (RSP4) which doubles
the performance of the RSP2
2. You all know about the VIP cards along with 11.2 when all of the bugs
are removed (especially those nasty memory leaks)
3. There has been a rumor that some level of load sharing between two
RSP's will be available some time 2Q or 3Q '97
Again, these are but rumors I've heard. We've been spending a lot of
our time lately playing with the Bay (Wellfleet) stuff, and they have some
pretty neat stuff coming out (thanks to ANS, any comments, Curt?)
1. They are working on a 64M version of the ARE card (dual Power PC
chips working in a full load share environment)
2. a 256M FRE3 card (dual Power PC's again) that they estimate would
be completely able to support as many as 300 BGP peers with 45K
3. Channelized T3 interface, native IP over non-traditional IP media, etc.
Again, mainly all rumors.
Chris A. Icide
More information about the NANOG