Re: (PM) Port Speeds

John W Baxter (jwblist@olympus.net)
Wed, 7 Jan 1998 13:15:53 -0800

>We are running PM2e30's with 33.6k modems on them. On all ports, we're
>running the speed at 115200. We're running both USR MP/16's and MicroCom
>ISPorte modems. I do not believe we are using hardware compression on the
>modems. Would there be an advantage in stability and/or performance if I
>dropped the speed on all ports to 57600? I ask about stability because I'm
>trying to debug a disconnect during download problem. I ask about
>performance, because I wonder if the portmaster would perform better
>running its ports at half the speed, and I remember a debate a while back
>about certain terminal servers not being able to deliver full bandwidth to
>all ports at 115200 (I could be wrong).
>
>Does anyone have any thoughts?

I would, as a customer, be unhappy with your lack of compression on the
modems, if true, and with a port speed less than 115200.

I routinely (at least twice a month) move highly-compressible large raw
text files over 28.8 connections at speeds which are limited by the port
speeds of 115200 at my end and the PortMaster end. When I happen to hit
one of our two modems which can't do 115200 for this task, I really notice
it.

[Really highly compressible: StuffIt reports 94% to 96% compression on the
files (which I *hope* means 4% to 6% of uncompressed size, but it's
somewhat meaningless as reported, like "97% fat free")...gzip only reduced
the most recent one by about half <just checked>.]

The transfers from modem to PortMaster and back tend to burst...they're
only likely to burst all at once when the primary backer of your company is
watching while you boast of how reliable the boxes are, and I suspect the
PM2 would survive even that.

--John (there's no such thing as an inanimate machine) Baxter

--
John Baxter (Born before ENIAC, but not by much.)
   jwblist@olympus.net      Port Ludlow, WA, USA

- To unsubscribe, email 'majordomo@livingston.com' with 'unsubscribe portmaster-users' in the body of the message.