I was chatting with someone the other day and we were trying to build a complete list of all units which can handle full routing tables 1 year from now, assuming current 4k/month growth (nevermind de-aggregation)
Juniper M/T-series units could handle 600k before, now 1mil with I-chip upgrade?
Juniper MX-series units are always 1mil
Cisco 6500/7600 with SUP720-3BXL handles 1mil routes
Force10 E300/600/1200 with dual-cam line cards handle 512k routes
Force10 E600/1200 with Exascale (quad-cam) line cards handle 1mil routes
Is there anything I'm forgetting here?
And if you already have one of these units, the upgrades are:
Juniper M-series units can replace the FPIC card to get new I-chip?
...if I understand it, no other cards need replaced
Cisco 6500/7600 you replace SUP32 or SUP720 with SUP720-3BXL
...if I understand it, no other cards need replaced?
(note that this disagrees with my understanding of how their FIB/CEF works so I'm curious about this)
Force10 you replace every single line card, since the entire chassis is limited to the smallest CAM size available.
The Foundry XMR series can also handle 1M routes in hardware and the CAM on all line cards is also appropriately sized. We have a mix of Sup720-3B-XL in 65XXs and Foundry XMRs. We have been happy with both and both should continue to serve us well into the future. The CAM can be partitioned into only IPv4, only IPv6, MPLS or a combination of all three.
-Robert
Tellurian Networks - A Perot Systems Company http://www.tellurian.com | 888-TELLURIAN | 973-300-9211
"Well done is better than well said." - Benjamin Franklin
Cisco 6500/7600 you replace SUP32 or SUP720 with SUP720-3BXL
...if I understand it, no other cards need replaced?
(note that this disagrees with my understanding of how their FIB/CEF
works so I'm curious about this)
If you have linecard DFCs they would need to be XLs also.
Keep in mind, on that platform, IPv4 and IPv6 routes share (rob from each other) space. 1mil IPv4 routes assumes you're not doing IPv6 at all. More realistic is some kind of split. i.e.
Cisco 6500/7600 with SUP720-3BXL handles 1mil routes
Sounds great on paper but a sup720 can barely handle full tables today.
Depending on how many full tables you take and what else you are doing
with it, cpu resources are unreasonably tight. Having many vlans with
vrrp and snmp polling also adds significant cpu load.
Also, beware the memory consequences of 'maximum-paths' in bgp
context. 8 full tables from a transit provider with maximum-paths=8
will exceed available ram on a sup720. With 6 you will have ~128m free.
Fortunately this is not a common configuration.
The rsp720 is substantially better at both of these issues. However the
rsp720 is only supported in 76xx chassis (officially) so chassis
selection is important for future upgrades.
An Alcatel 7750SR can support over 1 million BGP routes in its FIB and I
assume that the Cisco XR12000 family would also be able to handle the full
table a year from now.
What about Cisco's ASR series? We're going to be turning up a
multihomed connection with two full BGP views and I think our reseller
is going to be recommending ASR series routers...
DFCs allow hardware forwarding on the linecard. (Otherwise it's a centralized lookup on the PFC.. if you do not exceed the PPS limits of the PFC, you may not need a DFC for your environment).
It's the same EARL as is on the PFC. The system will use the lowest rev PFC/DFC type and they must all be the same type major class. eg: PFC2 series vs PFC3 series.. and it will follow the 3C[-XL] -> 3B[-XL] -> 3A downrev path. If you have a sup720 w/ pfc3a the dfc-3bxl's will lower themselves to 3a capabilities.
btw, this discussion is likely best suited for cisco-nsp.
Cisco 6500/7600 you replace SUP32 or SUP720 with SUP720-3BXL
...if I understand it, no other cards need replaced?
(note that this disagrees with my understanding of how their
FIB/CEF
works so I'm curious about this)
If you have linecard DFCs they would need to be XLs also.
The DFCs are what make hardware forwarding possible, yeah?
On the line card.
Upgrading the DFC requires a line-card swap, or just a DFC daughter-card?
You remove the standard CFC daughter card and put the DFC instead.
I believe the Cisco ASR series can handle 1M IPv4 or 250K IPv6 in
their entry level unit (ASR1002), the more advanced units can be
upgraded to handle 4M IPv4 or 2M IPv6. It does appear to be a "shared
table" design, so you would have to chose one or the other.