🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.dcom.lans.misc
Article #199

Re: T1 and LAN performance..??

#199
From: Rich Seifert
Date: Sun, 27 Sep 1992 16:52
51 lines
2838 bytes
In article <1992Sep24.204418.936@mdd.comm.mot.com>, pchan@mdd.comm.mot.com (Pat Chan) writes:
> I've two Ethernet LAN's (10Mbps) connected together through 
> bridges and dedicated T1 (~1.5Mbps), as follows:
> 
>        LAN A           T1          LAN B
>     --------- Bridge ------ Bridge------------
> 
> I've a file in LAN A. Say an application (in LAN A) requires 
> w seconds to retrieve this file. 
> Does it mean that an application requires w*10/1.5 seconds 
> to retrieve the file from LAN B????

This is a wonderful question!! It forces us to look at what is 
really going on.

In the initial case (file transfer on a single LAN segment), the
length of time to transfer the file is a function of:
   -The performance of the application responsible for the file transfer
    (e.g., the FTP implementation)
   -The transfer rates of the disks themselves on the two machines
   -The protocol overhead in the two stations (IP, other NOS services)
   -The device drivers for the NIC cards
   -The NIC cards themselves
   -and finally the tranfer rate of the Ethernet.

As may be becoming obvious, in most situations, the Ethernet is the
least of your bottlenecks. I know of very few applications (and file
transfers with normal NOS's are not among them!) which can keep
an Ethernet busy. (NFS diskless workstations and DEC Local Area VAXclusters
are exceptions.)

Soooo, when we move the file to the other side of this WAN bridge, we
add a few things to the delay:
   -The store-and-forward time of the two bridges
   -Queueing delay in each bridge (dependent on offered load)
   -transmission time on the intervening link

The transmission delay on the T1 is 10/1.5 longer than than on the
Ethernet, but queueing delay probably dominates, unless traffic is
quite light. In any case, the transmission delay is probably still 
much less than the processing delays encountered in the original case.

The good news is that, while performance will be worse in the second case,
it is not be a factor of 10:1.5   The bad news is that the performance
was probably lousy in the first case.

-- 
Rich Seifert                    Networks and Communications Consulting
seifert@netcom.com              (408) 996-0922
                                (408) 996-2860 FAX
"... specialists in Local Area Networks and Data Communications systems"
Technical Details
Message-ID: <5r#pp_g.seifert@netcom.com>
Path: arc1-ca.pugleaf.net!sparky!uunet!ogicse!usenet.coe.montana.edu!decwrl!csus.edu!netcom.com!seifert
References: <1992Sep24.204418.936@mdd.comm.mot.com>

Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!ogicse!usenet.coe.montana.edu!decwrl!csus.edu!netcom.com!seifert
From: seifert@netcom.com (Rich Seifert)
Newsgroups: comp.dcom.lans.misc
Subject: Re: T1 and LAN performance..??
Message-ID: <5r#pp_g.seifert@netcom.com>
Date: 27 Sep 92 16:52:19 GMT
Article-I.D.: netcom.5r#pp_g.seifert
References: <1992Sep24.204418.936@mdd.comm.mot.com>
Organization: Netcom - Online Communication Services  (408 241-9760 guest)
Lines: 51