On 21/03/2012 11:58, K wrote: > 1. How many 'I/O slots' does ASR-9010 have? 8 or 10?
8 slots. ref: > http://www.cisco.com/en/US/prod/collateral/routers/ps9853/data_sheet_c78-501767.html#wp9000166 > 2. Are RSP slots combo slots (can they be used to host I/O cards)? Not that I'm aware of. > 3. Does -L linecard (out of -L, -B and -E) support H-QoS (Hierarchical > QoS)? Cisco documentation is not clear Yes, it does. However it has significantly smaller buffers than the other two cards so if you're planning to do very rich hqos stuff they may not be the best choice. > 4. What is the difference between RSP440-SE and RSP440-TR? afaik, the only difference is that RSP440-SE has twice the route-processor RAM that the RSP440-TR does (12G vs 6G). This affects scaling for multiservice edge / carrier ethernet / larger scale L3 configurations. > 1. Do both provide a fabric throughput to each line-card at 440Gbps > now? They each provide 200G capacity to the line card edge ports. > 2. What is the actual throughput of these RSP440-SE/TR per line-card? This is a difficult question. ASR9K boxes haven't been dissected by users in the same way that e.g. sup720/rsp720/etc boxes have been. So while it's nominally 200G per line card, it hasn't been trashed around enough in real life for people to get any idea what its limitations are. > 5. What is the throughput of A9K-RSP-4G/8G per line-card? > 6. A9K-4T or A9K-8T provide 4/8 x 10GE ports at line-rate respectively? > 7. A9K-8T/4 provides only 4 x 10GE ports at line-rate? (Uses > over-subscription)? The short answer is that they each support line-rate. The longer answer is that it depends on what you measure, how you measure it and what you're doing on all the ports. There's a good description of how this all works here: > http://www.cisco.com/en/US/docs/routers/asr9000/hardware/overview/guide/asr9kOVRGbook.pdf See chapter 2. All of this is worth reading if you're planning to look at asr9k boxes. So with an RSP2, you get up to 80G usable bandwidth per slot. This is provisioned using N * 23G paths from the line-card fabric interface to the back-plane (where N = 2 for the 4x10G cards and 4 for the 8x10G cards). For each of those 2x23G paths, there is a 30G capable ASIC which interfaces to two NPUs. Each ethernet interface has its own dedicated NPU which can handle up to 15G traffic. So the entire system is nominally under-subscribed from each ethernet port right back to the backplane. This is necessary to deal with various internal framing overhead issues (which is one of the primary reasons that the c6500/sup720 start dropping packets well before you hit 40gigs). An NPU is a software forwarding engine. The more complex the configuration, the slower it operates. You'll certainly get line-rate with a simple in/out ipv4 config. But if you jam on ridiculously large ACLs with a crazy hqos config with tiny packet sizes and so forth, you may find that it's not going to be able to handle line rate any more. This is one of those "ymmv" issues. > 8. Do the above 10GE line-cards support DWDM-XFP-C= on each of the 10G > ports? (I need tunable DWDM XFP) yes, ref: > http://www.cisco.com/en/US/docs/interfaces_modules/transceiver_modules/compatibility/matrix/OL_6974.html#wp56063 If you need FEC, you will also need a A9K-ADV-OPTIC-LIC per line-card where you intend to run it. This isn't necessary for most installations and you'll really only need it for longer distances. > 9. Do I need a license per line-card for (A9K-AIP-LIC-B) for full-scale > L3VPNs? > 1. Is there a chassis-wide license? yes, you need a per-linecard license. There is no chassis-wide license. I understand that Cisco treat this as a feature rather than a bug. > 10. Does Cisco support L2VPN Draft-kompella on IOS-XR now? no idea. Nick _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
