cholla Posted May 29, 2005 CID Share Posted May 29, 2005 Hi all: I put these in another post but I wanted to put them in their own topic. Does anyone know why the speedguide.net Tcp/IP analyzer multiplies MSS by 44 to calculate RWIN ? You have to have scaling enabled for everything but the smallest RWIN value in each list. These are the suggested RWIN's from the speedguide.net Tcp/IP analyzer The first one is for PPPoE with 1492=MTU 1452=MSS 511104 (MSS x 44 * scale factor of 8 ) 255552 (MSS x 44 * scale factor of 4) 127776 (MSS x 44 * scale factor of 2) 63888 (MSS x 44) The second one is for other broadband 1500=MTU 1460=MSS 513920 (MSS x 44 * scale factor of 8 ) 256960 (MSS x 44 * scale factor of 4) 128480 (MSS x 44 * scale factor of 2) 64240 (MSS x 44) The third is for PPPoE using XP & without a router. 1480=MTU 1440=MSS 506880 (MSS x 44 * scale factor of 8 ) 253440 (MSS x 44 * scale factor of 4 126720 (MSS x 44 * scale factor of 2) 63360 (MSS x 44) Quote Link to comment Share on other sites More sharing options...
cholla Posted May 31, 2005 Author CID Share Posted May 31, 2005 Hey all; I didn't get any replies to this so I'm double posting it.I think the reason they use 44 X MSS is it comes closest to 65535 on the unscaled RWIN or the lowest one the number is the same. Anyone have a different or better explanation? Quote Link to comment Share on other sites More sharing options...
cak46 Posted May 31, 2005 CID Share Posted May 31, 2005 Just a guess, but could it be a function of the way ethernet works, at the firmware, hardware layer? Has to be a standard of some sort. If I remember correctly, setting RWIN and mtu correctly alleviates fragmenting of packets. Cholla: Please correct me if I'm wrong. My response is a long shot, but what the hay.... Edit: Darn: Spelling AGAIN also replaced mss with RWIN Quote Link to comment Share on other sites More sharing options...
cholla Posted May 31, 2005 Author CID Share Posted May 31, 2005 cak46; The MSS & MTU set at the correct size transfer without having to be broken down into smaller packets & refragmented.The settings inthe first post are supposed to be correct for that connection.If you multiply 1440,1452 or 1460 by 44 all are less than 65535 which unless I don't rember it correctky is the maximum Windows can receive unscaled.So I thought multiplying by 44 was as close as each one could get without going over 65536.for example 46 X1440=66240. The multiplier has to be an even number or scaling doesn't work correctly. Quote Link to comment Share on other sites More sharing options...
cak46 Posted May 31, 2005 CID Share Posted May 31, 2005 fiddlin around I used an RWIN of (MSS x 44 * scale factor of 16 ). Seemed to work ok. Another question that comes into play for those mathematically inclined is why the scale factor being a multiple of 2 (?) or exponential of 2? I get huge acks when my rwin is small....... Quote Link to comment Share on other sites More sharing options...
cholla Posted May 31, 2005 Author CID Share Posted May 31, 2005 I think if your RWIN is too small for the data coming in it temporaily doesn't have anywhere to go.So it returns until it can load.I think thats what causes the acks. Quote Link to comment Share on other sites More sharing options...
cak46 Posted May 31, 2005 CID Share Posted May 31, 2005 I think if your RWIN is too small for the data coming in it temporaily doesn't have anywhere to go.So it returns until it can load.I think thats what causes the acks. That would explain my problem with not using scaling. Watched it with ethereal one night.. and got ACKs up the ying yang at 65535 Quote Link to comment Share on other sites More sharing options...
cholla Posted May 31, 2005 Author CID Share Posted May 31, 2005 cak46 : what do you use to tell how many ACKS you are getting? Quote Link to comment Share on other sites More sharing options...
cak46 Posted May 31, 2005 CID Share Posted May 31, 2005 Ethereal..... It's a packet sniffer. Prior to starting the test I started ethereal then ran the test then stopped ethereal. Heres a link to it, but you need another program to go along with it... WinPcap. This is the link for ethereal: www.ethereal.com It can give you some insight on what really occurs with packet transmissions and the information contained therein. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.