lrm@b.cs.wvu.wvnet.edu (GTA Systems) (01/19/89)
I would like to know how fast AppleTalk and AppleShare with CAP 5.0
should run with the following setup:
Three appletalk zones:
The first two are appletalk networks connected to our
ethernet backbone. The bridges are two Kinetics Fastpath
boxes running KIP0688.
The third zone is the Ether Talk zone and it is used to
route to our Macintosh II's with Kinetics Ether Port II
cards.
We have full functionality, but I have a few questions.
First, In order for the Mac II's to register and boot with system
software 6.0.2, I was forced to use a different zone name for
the Ethertalk line in the atalkatab. Is this normal? If you
need more info to answer please let me know.
Second, The speed of appleshare on the appletalk zones is slow, but
I don't feal it is too bad for appletalk. The appleshare speed
of the Mac II's with the Kinetics EtherPort II is even SLOWER
than that on the appletalk. Why? Is all appleshare data being
restructured by the fastpath box with the ethertalk address in
the atalkatab and then being sent to the UNIX file servers?
I don't know how KIP and CAP handle this and I would like to know
for both work and curiosity. If you need more info let me know.
Third, is it possible to ftp from one MAC to another. I have
been using BYU's updated version (2.2.3) of NCSA telnet that has
a "full" implementation of ftp, ie. I can both ftp to a host
and ftp back to my Mac from a host. But, I cannot ftp from
one Mac to another or from a host to a different Mac with telnet
up and running. Is it possible that it is looks like it is
looking out on the ethernet for addresses, but it is actually
only connecting to the Mac because the connection has already
been established by the original telnet access i.e ftp would need
to invoke NBP?
Fourth, Why does the non-ethertalk kinetics box show up on
my ethertalk zone when I do a atlook for only the ethertalk
zone? Is this normal?
I really think we have a setup bug in KIP, but as I said we have
full functionality with CAP and NCSA/BYU Telnet. The really
good point to the KINETICS EtherPort II is that when I set
mtu and maxseg to 1024 and rwin to its max setting for ethertalk.
I can transfer a 321061 byte file from a SUN 4 at 24 Kbytes/s in
13 seconds and I can send it back to the SUN 4 at 19 Kbytes/s in
17 seconds on an empty system. But I have noticed that BYU's
implementation of ftp for the same transfer was significatly
slower, making me believe that they are using my login info
to login with telnet and facking the login as ftp, could this be
possible? On second thought, their implementation must be really
poor.
All help in tweeking appletalk and appleshare would really be
greatly appreciated by me and my department. We would really
like to serve files to our etherport II macs much faster.
If you need more info or just have a reply, please mail to:
lrm@a.cs.wvu.wvnet.edu
or post
Thanks your time.
Larry Metzger
GTA, Systems
West Virginia University