Frame-Relay encapsulation on MPLS, throughput question

[robert.mcclain]

I have 2 - 3925 routers running 15.2(4)m4 code. I have a full DS-3 connected to the serial interface. Running Frame-relay encapsulation over an MPLS cloud provided by a carrier. Only one client utilizing this circuit. When they run a SCP from server to server the throughput is horrible. The copy starts out at around 2M a second but quickly drops down to 700-800kps. Ping times between the two sites is around 63ms on average. We have tried several options on the servers to increase the throughput, including increased window size (256k) but it doesn't change the outcome. Is this a limitation due to the Frame-Relay encapsulation? If so is there a way to change it or do I need to change the encapsulation to PPP?

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

I doubt it's the frame-relay encapsulation.

Are you sure your provider isn't doing any rate-limiting?

Ok, one side is a full DS-3, and the other side?

Is your topology only logically p2p or it is multipoint?  If the latter, is there any other traffic from other nodes?

[robert.mcclain]

Joseph,

It is a full DS-3 on both sides, multipoint but there are no other sites on the cloud.

Provider states they aren't rate limiting.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Well, then I would suggest some off-time transfer rate testing.  You should, from what you describe, be able to push about 45 Mbps in either direction.  If not, then your SP can explain why not.

[robert.mcclain]

My first thought was they were rate limiting us, so I got them on the phone and they looked and determined they weren't. then later on I get them back on a troubleshooting call with the people who were testing the throughput and the Sp came back with increasing the send/receive buffers on the AIX servers. So that was done, went from 64k to 256K. No change in the throughput.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

SP do make mistakes too.  Which is one reason you need to test the bandwidth.  (NB: you want to test with bandwidth generator, so you can control test rates.)  You need to verify you can push 45 Mbps, in either direction, without drops.