Buffers and TCPWindowsize...WAS: PTP latency
Dear list, Thank you for all the great responses regarding the PTP latency issue. This was very thorough information that will help me make a decision. A related, yet somewhat separate issue that I would like some advice on is what kind of circumstances would cause a tcp congestion window to be reduced over a WAN connection, assuming that the servers on both sides are not busy (or even barely utilized). Assuming that a 1472 byte packet could pass freely from WAN point A to WAN point B without being defragmented, it has been suggested to me that somehow windowing is playing a role in poor throughput. I understand the role that windowing plays in tcp/ip, I guess what I'm wondering is how that packet could get 'squeezed' as it passes through a WAN. Thank you, Christopher
On Mon, 01 Oct 2001 21:24:11 PDT, Christopher Wolff <chris@bblabs.com> said:
I understand the role that windowing plays in tcp/ip, I guess what I'm wondering is how that packet could get 'squeezed' as it passes through a WAN.
Is it possible that you need RFC1323 extensions to get the expected throughput due to delay-bandwidth product, but one end or the other isn't doing 1323? /Valdis
participants (2)
-
Christopher Wolff
-
Valdis.Kletnieks@vt.edu