[AusNOG] Egress shaping on Telstra EA Service

[Andrew Yager]

Hi,

I'm working through troubleshooting a 200M EA service. Shapers are done on
Juniper MX (head end) and Juniper SRX300 (customer end).

Ingress we are receiving full bandwidth to shaping value (195Mbps)
Egress we are only receiving 30Mbps for TCP traffic only.

If I clamp the shaper down to 10Mbps, I see that the speed drops to a flat
9.8Mbps.

If I do an iperf using UDP, I'm seeing the full 195Mbps from client machine
out.

Does anyone have any ideas about what this could be? We've got a few other
SRX300s in the field using the same config; and so it doesn't look like
that's the problem - and also given that clamping the shaping to 10Mbps
evens the results it seems like the SRX won't be what is at fault.

Shaper config on customer side:

# show class-of-service
interfaces {
    ge-0/0/0 { <!-- this is facing customer machine for testing
        unit 0 {
            scheduler-map policer;
        }
    }
    ge-0/0/5 { <!-- this is facing Telstra NTU
        unit 0 {
            scheduler-map policer;
        }
    }
}
scheduler-maps {
    policer {
        forwarding-class best-effort scheduler rate-limit-195m;
    }
}
schedulers {
    rate-limit-195m {
        shaping-rate 195m;
    }
}

# run show class-of-service scheduler-map policer
Scheduler map: policer, Index: 55355

  Scheduler: rate-limit-195m, Forwarding class: best-effort, Index: 19467
    Transmit rate: unspecified, Rate Limit: none, Buffer size: remainder,
Buffer Limit: none, Priority: low
    Excess Priority: unspecified
    Shaping rate: 195000000 bps
    Drop profiles:
      Loss priority   Protocol    Index    Name
      Low             any             1    <default-drop-profile>
      Medium low      any             1    <default-drop-profile>
      Medium high     any             1    <default-drop-profile>
      High            any             1    <default-drop-profile>


UDP Iperf from client outbound that shows it's not the link capacity…

[ ID] Interval      Transfer    Bandwidth        Jitter  Lost/Total Datagrams
[  3]  0.0-27.7 sec  636 MBytes  192 Mbits/sec  0.029 ms 24291/478152 (5.1%)
[  3]  0.0-27.7 sec  1 datagrams received out-of-order
[  4] local xx port 5001 connected with xx port 58752
[  4]  0.0-10.3 sec  229 MBytes  187 Mbits/sec  16.141 ms 8958/172410 (5.2%)
[  4]  0.0-10.3 sec  1 datagrams received out-of-order
[  3] local xx port 5001 connected with xx port 60477
[  3]  0.0-10.0 sec  228 MBytes  191 Mbits/sec  0.039 ms 9595/172190 (5.6%)
[  3]  0.0-10.0 sec  1 datagrams received out-of-order
[  4] local xx port 5001 connected with xx port 50602
[  4]  0.0-30.0 sec  683 MBytes  191 Mbits/sec  0.023 ms 29513/516460 (5.7%)


Any advice/input would be appreciated. I'm also curious if there is any
chance this could still be Telstra?

Thanks,
Andrew
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20161216/2ed08616/attachment.html>