NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/dcom/lans/misc/949

Text File | 1992-11-20 | 5.5 KB | 116 lines

Newsgroups: comp.dcom.lans.misc
Path: sparky!uunet!portal!kalpana!johnb
From: [e-mail] (John Bartas)
Subject: Re: Token Ring vs. Ethernet
Message-ID: <[e-mail]>
Organization: Kalpana Inc
References: <[e-mail]> <[e-mail]> <[e-mail]>
Date: Sat, 21 Nov 92 03:48:33 GMT
Lines: 105

I just have to jump into this fray  :->

In article <[e-mail]> [e-mail] (Richard Brett Frankenberger) writes:
>In article <[e-mail]> [e-mail] (Vernon Schryver) writes:
>>-Ethernet is still cheaper, and that is not likely ever to change.
>
>True.
>

MUCH cheaper! A few days ago I bought a first class 16 bit for $170.00, and
thats without shopping around (I was in a hurry). I've seen PC
ethernet adapters advertised for under $100. The lowest priced token Ring card
I've seen was around $400.00 (that was about a year ago). Even allowing
for some price drop, the token adapters are still over twice what
the ethernet cards are.

>>-Ethernet is simpler and so more robust.  Any single ring scheme is a
>> royal pain compared to ethernet.  The hassles of source routing and
>> other token ring "features" must be experienced to be believed.
>
>It's more robust?  Really?  What happens if a faulty transciever cable in
>ethernet is messing up the entire line?  Nothing.  On Token-Ring?  The
>workstation will remove itself from the ring.  I can bring down an ethernet a
>lat easier than I can bring down a token-ring.  And I can get a token-ring back
>up faster than an Ethernet.
>

In theory, token rings should be almost indestructable, but the small 3-4
node ones I've set up in labs always seem to find ways to go belly-up.
The token rings I've used within the last year seem to be about as reliable
as ethernets I used in the mid '80s. The Ethernets have improved a lot since
then. I have no empirical data to support this, but I'm convinced it's
so just based on personal experience.

>>-10MHz Ethernet is faster than 4Mb/s token ring.  A pair of 16Mb/s token
>> ring stations can be 60% faster than a pair of ethernet stations,
>> but mostly because of the underlying bandwidth.  At the (now) low
>> speeds of ethernet and token ring, there is no interesting effeciency
>> difference for stations (not hubs or switches) between packet sizes
>> of 1500 and 16K.
>
>Really?  On a 4 MBPs token-ring, I ran some large file copies at approx 1.5K and
>4K packet sizes.  The 4K was about 25-50% faster (I can't remember exactly).
>Of course, for smaller transactions, the difference is less, but still
>noticable.  (This test was done with a lightly loaded Novell file server with
>Packet Bursting (without packet bursting, the difference would be even more
>pronounced).
>

About a year ago I used Lantest (public domain PC adapter test utility) to
test Token Ring and Ethernet adapters for relative speed at different
frame sizes. I used the NDIS driver interface for for types of media.
The measureable speeds, both pps & Bps, seemed to depend on the
NDIS driver as much as the packet size of media type. My recollection
is that 10Mbit ethernet on a good 16 bit card with a good driver at 1514
byte packets outran the 16Mbit token ring card (8 bit bus interface)
using 4K packets. I think the 8bit bus interface was more of an issue than
the packet size.

    I won't post thoses numbers, because I had contracts with some of the
manufacturers, (& they are about a year old), but I have an idea:

If there's sufficient interest I'll see if I can borrow some
newer hardware and stage Ether vs token races in my lab some weekend.
I have lots of new ethernet hardware, but all my token ring hardware is
obsolete or engineering prototypes. Can anyone in the San Jose area
bring some pairs of popular Token Ring cards (I have some older 8 bit
IBMs)? It would help if we had several volunteers present to assure
impartial testing and results.  We could perform a standard
set of tests using Lantest at various packet sizes and hardware
configurations and publish a scorcard of the results.

As I think about this, it seems someone must have done this. Does anybody know
of such a scorecard? Any interest in helping to start one?

>Since someone is sure to bring this up, at heavy loads, ethernet's performance
>is much worse than token-ring (too many collisions).  At very light loads,
>Ethernet's performance is better (no waiting for tokens).  (I'm comparing


I seem to recall a bug in a well know token ring MAC software package
that caused the PC adapter to lock up if the ring got too loaded. Again,
the maturity of the Ethernet products made them more functional than the
theoreticlly superior token ring adapters.

>the topologies here independant of speed.  Of course 10MBPs is faster than
>4MBpsand 16MBps is faster than 10MBps), butthe token-wait time is

    As my comments above indicate, I don't believe that 16 Mbs Token
will always give the end user more speed or throughput that 10 Mbs
Ethernet.

    I've accidently deleted the rest of the original posting, but
I seem to recall some thing about FDDI being an upgrade path for
4/16 Mbit token ring. Soon we will have 100 Mbit ehternet, which should be
a fairly painless upgrade for ethernet users.

Cheers,

======================================================================
John Bartas                     | Disclaimer - I don't speak for Kalpana
Pricipal Software Engineer      | in any official capacity. I'm just
Kalpana, Inc. [phone]x141 | an employee who reads the net.

"We have met the enemy, and he is us." -Pogo