Rwin

chaobox99 · Post by **chaobox99** » Tue Aug 21, 2001 12:56 pm

I just read about the Cablenut Dictator words in broadband nut forum. The "correct" RWIN should be approximately

RWIN = max. bandwidth x average RTT

If this calculation is correct, is that mean we don't have to try any other values?

So, how does this calculated RWIN compare to the default 64240,128480,256960,513920?

Lobo · Post by **Lobo** » Tue Aug 21, 2001 1:05 pm

Very easy, use what your system likes best

chaobox99 · Post by **chaobox99** » Tue Aug 21, 2001 1:19 pm

I think the system is a box, everything is linear, isn't it?

There should be an equation to pinpoint an optimum RWIN value that will give you the optimum performance. May be I just think like a box.......

Does anyone know if there are any other parameters for RWIN beside bandwidth & RTT?

And, can someone tell me what "btw" means

, it's been pondering me for quite a while, I can never figure it out....

Lobo · Post by **Lobo** » Tue Aug 21, 2001 1:34 pm

By the Way

cablenut · Post by **cablenut** » Tue Aug 21, 2001 4:38 pm

Your RWIN is a total factor of your Connection Speed & Latency across your connection.

RWIN Discussion

The end users system really doesn't matter per say, as long it is optimal enough to accept the Internet connection to 100% it should be fine.

rmrucker · Post by **rmrucker** » Tue Aug 21, 2001 4:46 pm

I am not sure if "average" RTT is what you want. I would agrue for "the highest RTT that you are likely to encounter on a recurrent basis". (Yes, I am aware that sounds stupid)!

For example, if you set your RWIN at your Maximum bandwidth * average RTT -- then you are limiting your window size unnecessarily. What if you connect to a site with a high latency/high RTT -- yet the bandwidth is also high??

Then your window will NOT be full during most of the transmission. A larger RWIN would allow you to maintain a full window during the entire transmission, and therefore usually a faster transmission.

Unless someone can convince you that the RWIN is "too high" when you use the "highest likely RTT", then I would recommend a RWIN that is GREATER than (Maximum bandwidth * average RTT).

I would use (Maximum bandwidth * "highest likely RTT") as my minimal RWIN. But you will find many opinions...

cablenut · Post by **cablenut** » Tue Aug 21, 2001 4:59 pm

Yes rmrucker I agree that is why in my post on my discussion board. I picked places where I knew I had many hops too (i.e. german sites, and canadian sites) I then took my Max latency and my Min from that group of sites and averaged them out. I wouldn't suggest finding your RTT from sites where you have little hops too, or your lowest latency too.