TCP congestion control kernel parameters

[Stephen F. Cawley]

I have a network application which needs to communicate in real time to clients which are spread around the globe.  There are times when there is not traffic on the network and the Initial Window shrinks.  When data is sent after these idle times, we are seeing up to a 1 second latency before the protocol comes up to speed.

Is it possible to avoid the TCP slow start algorithm and send at full speed or must we step up with every ACK?  I have been looking at the tcp_wmem and tcp_rmem kernel parameters, but am unsure whether they are the ones that I need to work with.

We are operating in a private network with RedHat 
2.4.9-e.25enterprise kernel.  

Thanks in advance.


--
Stephen F. Cawley




_______________________________________________
No banners. No pop-ups. No kidding.
Introducing My Way - http://www.myway.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/