24e12c4b
From: free@securityaudit.val.newsbank.net
In article ,
chris writes:
> I've got TCP performance problem with a Windows 2003 box, hope
> somebody can advise me on the tuning parameter I'm looking for.
>
> The application stops and starts its transmissions, pausing for ~10
> seconds between long periods of bulk throughput.
>
> When the application is actually sending, things go well: Data rates
> are acceptable, and I can clearly see TCP expanding the transmit
> window one segment at a time. The transmission rate, and gap between
> segments is quite smooth and steady during these intervals. In-flight
> data tends to peak around 100KB. Packets are not frequently lost
> during steady-state operation.
>
> Then transmission then stops abrubtly. After some time, the
> transmission restarts, just as abruptly as it stopped. This is where
> things get ugly. Instead of these segments being spaced out nicely,
> the whole congestion window is dumped onto the wire at the server's
> line rate (100Mb/s).
See section 4.1 of rfc2581 and rfc5681:
Therefore, a TCP SHOULD set cwnd to no more than RW before beginning
transmission if the TCP has not sent data in an interval exceeding
the retransmission timeout.
> These bursts tend to be 16 segments (just under 24KB)
The rfcs prescribe min(IW, cwnd), where
IW = min (4*SMSS, max (2*SMSS, 4380 bytes))
and:
a TCP sender MAY use a 3 or 4 segment initial
window, provided the combined size of the segments does not exceed
4380 bytes.
rfc2581 is ten years old, and Van Jacobson recommended this 20
years ago. I find it hard to believe that ms-windows does not do this
out of the box.
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
