<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Pete, these are production boxes with
multiple flows passing through.<br>
<pre class="moz-signature" cols="90">--
Tassos
</pre>
Pete Lumbis wrote on 2/2/2013 22:11:<br>
</div>
<blockquote
cite="mid:CAB0xJrM500Ji62hP_7hURRZPkP6FhBC8L4DxAq2mrwi4RoyrvA@mail.gmail.com"
type="cite">
<div dir="ltr">How are you testing? Single flow or multiple flows?</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">On Sat, Feb 2, 2013 at 6:33 AM, Tassos
Chatzithomaoglou <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:achatz@forthnetgroup.gr" target="_blank">achatz@forthnetgroup.gr</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
I have the following setup on a ASR1006 (RP2/ESP40/SIP40)
and i'm trying to find out if<br>
the following behavior is expected.<br>
<br>
<br>
+-----+<br>
Te1/0/0 (4G/1G) ---| |---Te1/1/0 (2G/8G)<br>
| |<br>
Te1/2/0 (4G/1G) ---| |<br>
+-----+<br>
<br>
<br>
When the output of Te1/1/0 goes above 8G, RTT for packets
flowing from Te1/0/0 to Te1/2/0<br>
increases by 50-100ms.<br>
<br>
The same happens in the following scenario on another
ASR1006 (RP2/ESP20/SIP10); when the<br>
output of Te1/1/0 goes above 6G, RTT for packets flowing
from Te1/0/0to router's loopback<br>
increases by 50-100ms (RP is at ~30% all the time).<br>
<br>
+-----+<br>
Te1/0/0 (6G/1G) ---| |---Te1/1/0 (1G/6G)<br>
| |<br>
| |<br>
+-----+<br>
<br>
<br>
Most of the times, RTT increase is followed by packet loss<br>
<br>
This reminds me of HOL blocking, but i had the impression
this was applicable mostly to<br>
switches with small buffers.<br>
At the same time, the ESP sends thousands of flow control
signals to the SIP that it can't<br>
cope with this traffic rate.<br>
<br>
ASR1006#show platform hardware slot 1 serdes statistics<br>
>From Slot F0-Link A<br>
Pkts High: 1687702827 Low: 391241384970 Bad: 0
Dropped: 0<br>
Bytes High: 326483900940 Low: 291059939598187 Bad: 0
Dropped: 0<br>
Pkts Looped: 0 Error: 0<br>
Bytes Looped 0<br>
Qstat count: 0 Flow ctrl count: 40306518521
<===<br>
<br>
>From Slot F1-Link A<br>
Pkts High: 0 Low: 0 Bad: 0
Dropped: 0<br>
Bytes High: 0 Low: 0 Bad: 0
Dropped: 0<br>
Pkts Looped: 0 Error: 0<br>
Bytes Looped 0<br>
Qstat count: 0 Flow ctrl count: 80093<br>
<br>
-after 1 sec-<br>
<br>
ASR1006#show platform hardware slot 1 serdes statistics<br>
>From Slot F0-Link A<br>
Pkts High: 1687721691 Low: 391244370772 Bad: 0
Dropped: 0<br>
Bytes High: 326487553571 Low: 291062458884384 Bad: 0
Dropped: 0<br>
Pkts Looped: 0 Error: 0<br>
Bytes Looped 0<br>
Qstat count: 0 Flow ctrl count: 40307432319
<===<br>
<br>
>From Slot F1-Link A<br>
Pkts High: 0 Low: 0 Bad: 0
Dropped: 0<br>
Bytes High: 0 Low: 0 Bad: 0
Dropped: 0<br>
Pkts Looped: 0 Error: 0<br>
Bytes Looped 0<br>
Qstat count: 0 Flow ctrl count: 80094<br>
<br>
<br>
<br>
ASR1006#show platform hardware slot 1 plim status internal<br>
FCM Status<br>
XON/XOFF 0x0000000000000003<br>
ECC Status<br>
Data Path Config<br>
MaxBurst1 256, MaxBurst2 128, DataMaxT 32768<br>
Cal Length RX 0x0002, TX 0x0002<br>
Repetitions RX 0x0010, TX 0x0010<br>
Data Path Status<br>
RX in sync, TX in sync<br>
Spi4 Channel 0, Rx Channel Status Full, Tx Channel Status
Hungry <===<br>
Spi4 Channel 1, Rx Channel Status Starving, Tx Channel
Status Starving<br>
RX Pkts 391387121421 Bytes 285048619167994<br>
TX Pkts 393073127218 Bytes 291507927293959<br>
Hypertransport Status<br>
RX Pkts 0 Bytes 0<br>
TX Pkts 0 Bytes 0<br>
<br>
<br>
<br>
TAC is talking about microbursts (how unusual), and although
i can't measure 10G traffic<br>
per ms, QFP's 5-sec data doesn't agree with them.<br>
<br>
ASR1006#show platform hardware qfp active data utilization<br>
CPP 0: Subdev 0 5 secs 1 min 5 min
60 min<br>
Input: Priority (pps) 884 840 838
835<br>
(bps) 783344 750704 746360
725664<br>
Non-Priority (pps) 1291098 1282015 1261073
1265900<br>
(bps) 7544814904 7465944936 7322679592
7345058240<br>
Total (pps) 1291982 1282855 1261911
1266735<br>
(bps) 7545598248 7466695640 7323425952
7345783904<br>
Output: Priority (pps) 9065 9191 9184
8897<br>
(bps) 11485520 11659776 11730880
11357888<br>
Non-Priority (pps) 1281141 1271903 1251199
1256512<br>
(bps) 7560289312 7481701984 7338684976
7360568224<br>
Total (pps) 1290206 1281094 1260383
1265409<br>
(bps) 7571774832 7493361760 7350415856
7371926112<br>
Processing: Load (pct) 59 59 59
59<br>
<br>
<br>
<br>
At the same time different IOS releases give different
results (15.2(4)S2 is far worsethan<br>
15.1(3)S2) and i'm starting to believe that ASR1006 is
another hype scheduled to go down...<br>
<span class="HOEnZb"><font color="#888888"><br>
<br>
<br>
<br>
--<br>
Tassos<br>
<br>
_______________________________________________<br>
cisco-nsp mailing list <a moz-do-not-send="true"
href="mailto:cisco-nsp@puck.nether.net">cisco-nsp@puck.nether.net</a><br>
<a moz-do-not-send="true"
href="https://puck.nether.net/mailman/listinfo/cisco-nsp"
target="_blank">https://puck.nether.net/mailman/listinfo/cisco-nsp</a><br>
archive at <a moz-do-not-send="true"
href="http://puck.nether.net/pipermail/cisco-nsp/"
target="_blank">http://puck.nether.net/pipermail/cisco-nsp/</a><br>
</font></span></blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</body>
</html>