DS3 support on Cisco routers

[thisisshanky]

Cisco Software Advisor shows that even any router starting from 2600 can take the NM-1A-T3 module for terminating a DS3.

Any issues with running DS3 on a 2600XM? Can it handle DS3 at full 45 Mbps ?

[eric.fairfield]

No a 2600XM cannot handle a full DS-3. A 7200 will support a full DS-3. A 3745 might be able to but am not 100 percent sure.

[thisisshanky]

Eric,

Thank you for the info. So do you know the limit on, how much the DS3 can burst to, on a 2600 XM.

Any 2620XM, 2650XM, 2691 will do.

[kkalaycioglu]

Not sure about NM-1A-T3 but according to http://www.cisco.com/en/US/products/hw/modules/ps2797/products_data_sheet09186a008010fba2.html

NM-1T3/E3 module can support only subrate E3/T3 services on 26XX and fullrate services for 36XX, 37XX routers.

Regards.

[thisisshanky]

Yep, thanks for the reply!

I saw the link, right after I posted this...I was infact looking for what exact values can it be, when you say about sub rate ? 6 Mbps, 12 Mbps, 24 Mbps etc on the 2600XM router.

[kkalaycioglu]

Hi,

I'm planning to use E3 NM with 2600 and looking for the same answer for this.

Regards.

[tbaranski]

A theoretical answer:

A 2621 (rated at 25k pps*) can handle about 40Mbps of traffic with 1500 byte packets per some testing that I did. (CEF, no access lists, etc.) A 3620 (20-40k pps) does about 75Mbps per someone else's testing. The 2650XM is rated at 40k pps.

So I'd guess that a 2650XM can get fairly close to filling a full-duplex DS3. This assumes they're not doing something goofy with the DS3 module like programming it to limit itself to a certain bandwidth when it detects that it's in a certain router. The router <-> module interface itself could also be a limiting factor.

* All of Cisco's pps ratings are with 64 byte packets.

[thisisshanky]

Terry,

At last I am getting some numbers. Thanks for that. The particular solution I am looking for involves a point to point DS3 between two locations. Mostly SQL replications, file transfer, web server access are the kind of traffic that is going to traverse the link.

When you said 40 Mbps traffic for the 2621, was it a full duplex or half duplex value ?

[tbaranski]

It was a one-way (except for the ACKs) TCP transfer with TTCP (unfortunately I didn't have access to a SmartBits at the time). The 3620 test was also done with TTCP.

Tests like this are, of course, pretty optimal relative to real-world traffic patterns like what you're expecting here.