Skip to Content.
Sympa Menu

ndt-users - Overoptimistic estimates

Subject: ndt-users list created

List archive

Overoptimistic estimates


Chronological Thread 
  • From: Jon Kåre Hellan <>
  • To:
  • Subject: Overoptimistic estimates
  • Date: Mon, 13 Feb 2006 14:42:40 +0100

Hi,

I've placed ndt servers at various points in our network, but find that way too many links are reported as 10 Gbps.

Here's part of a web100srv logfile:

Web100srv (ver 3.1.4) process (9461) started Feb 13 14:36:51
Feb 13 14:36:53 persaunet.uninett.no port 49412
spds[0] = ' 0 0 300 159 2105 5281 53 6 30 7632 0 0 38.68' max=7632 [49.03%]
spds[1] = ' 0 0 300 73 767 3735 1516 0 338 1438 0 0 51.71' max=3735 [45.73%]
spds[2] = ' 0 0 37 221 1050 1116 337 33 948 1030 0 35 44.74' max=1116 [23.39%]
spds[3] = ' 0 0 81 666 3219 1987 345 1 0 4903 1 1147 73.74' max=4903 [43.77%]
Feb 13 14:36:53,158.38.62.138,32202,17823,0,413062,12352,0,0,26699,0,1448,1,12352,\
131072,8388608,1449448,998791,9378723,9880,38977546,29,30,1,0,233,131072,100,\
0,0,0,0,9,5,5,9,0,26699,31,7,22

Here's what the applet says:
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client to server) . . . . . 17.88Mb/s
running 10s inbound test (server to client) . . . . . . 29.74Mb/s
The slowest link in the end-to-end path is a 10 Gbps 10 Gigabit Ethernet/OC-192 subnet
Information: The receive buffer should be 40820.31 Kbytes to maximize throughput

And here's what analyze -d 5 says when run on this log:

Start of New Packet trace
Web100 variables line received

Client --> Server data detects link = 10 Gigabit Enet
Client <-- Server Ack's detect link = FastEthernet
Server --> Client data detects link = FastEthernet
Server <-- Client Ack's detect link = 10 Gigabit Enet
Acks = 0.4626, async speed = 0.5535
Throughput to host [persaunet.uninett.no] is limited by a '10 Gigabit Enet' subnet
Web100 says link = 100, speed-chk says link = 9
Speed-chk says {9, 5, 5, 9}, Running average = {38.7, 51.7, 44.7, 73.7}
C2Sspeed = 17.82 Mbps, S2Cspeed = 32.20 Mbps, Estimate = 33035.48 Mbps (33035.48 Mbps)
Old estimate is greater than measured; New estimate is greater than measured
Loss = 0.00% (0.00%), Out-of-Order = 0.01%, Long tail = {0, 2, 3, 0}
Distribution = {2, 1, 2, 3}, time spent {r=9.6% c=90.3% s=0.1%}
Ave(min) RTT = 33.44 (31) msec, Buffers = {r=131072, c=1449448, s=4194304}
bw*delay = {r=31.36, c=346.75, s=2006.79}, Transitions/sec = {r=2.9, c=3.0, s=0.1}
Retransmissions/sec = 0.0, Timeouts/sec = 0.0, SlowStart = 0
Mismatch = 0 (0:0[1.00]), Cable fault = 0, Congestion = 0, Duplex = 0

Version of ndt is ndt-3.1.4a. The Linux kernel is 2.6.12.2 with web100 patches. The path actually includes a 100 Mbps segment to my workstation, and a 155 Mbps subsea fiber from mainland Norway to Spitsbergen.

Any ideas?

Jon Hellan
UNINETT



Archive powered by MHonArc 2.6.16.

Top of Page