[comp.sys.mac.comm] Gatorbox v Fastpath

jf@ap.co.umist.ac.uk (John Forrest) (03/06/91)

Would anyone like to advise me on the various merits of
Gatorboxes and Fastpaths. I'm hoping we will get some money to
buy a couple, but it not clear which is the most appropriate.
The price here seems about the same (#1850+VAT), so some
clue as to which to buy would help.

John Forrest
Dept of Computation
UMIST

dorner@pequod.cso.uiuc.edu (Steve Dorner) (03/06/91)

In article <1991Mar5.220625.6493@cns.umist.ac.uk> jf@ap.co.umist.ac.uk (John Forrest) writes:
>Would anyone like to advise me on the various merits of
>Gatorboxes and Fastpaths. I'm hoping we will get some money to
>buy a couple, but it not clear which is the most appropriate.

I've dealt with both.  Here are the pros and cons I've run into:

FastPath PROS:
- doesn't need tickling after power outage
- configuration application is easier to use

FastPath CONS:
- the configuration itself is hard; lots of numbered options

GatorBox PROS:
- the configuration itself is understandable; English words used throughout
- NFS->AFP and LPD->PAP translations available at extra cost
  (if you have UNIX machines, you can use CAP with either box for similar
  functionality.)
- built-in support for IP tunnelling (gatorboxes can talk to each other
  over IP internets)
- very good phone *and email* support

GatorBox CONS:
- the configuration application is a wee bit weird
- requires reload after power failure (can be automated with tftp)

Both boxes seem to be pretty reliable in terms of hardware and software.

My contact with the boxes is mostly in their AppleTalk<->IP function,
only secondarily as AppleTalk routers.
--
Steve Dorner, U of Illinois Computing Services Office
Internet: s-dorner@uiuc.edu  UUCP: uunet!uiucuxc!uiuc.edu!s-dorner

jwk@Scripps.EDU (John Kupec) (03/07/91)

In article <1991Mar5.220625.6493@cns.umist.ac.uk> jf@ap.co.umist.ac.uk (John Forrest) writes:
>Would anyone like to advise me on the various merits of
>Gatorboxes and Fastpaths. I'm hoping we will get some money to
>buy a couple, but it not clear which is the most appropriate.
>The price here seems about the same (#1850+VAT), so some
>clue as to which to buy would help.
>
>John Forrest
>Dept of Computation
>UMIST


OK, my two cents on this:

First, both MacUser & Macworld have articles about LocalTalk-Ethernet
gateways/bridges in the April 91 issues.

I have configured/used both FastPaths and GatorBoxes.  The functionality
provided (GB easily, FP with work) is Telnet connections to unix 
hosts, using unix hosts as NFS fileservers, printing to unix machine 
printers, printing to Mac laserwriters.

The FastPaths are used in conjunction with CAP running on a Sun 3/60.
The GatorBoxes run GatorShare which includes GatorPrint.

If you want all of the above features then the GatorBox is the hands-down 
choice between the two.  It's easier to configure, maintain, and (IMHO) has 
a better future.   A GatorBox with GatorShare software would be your ticket.

If all you want is TCP/IP connectivity for Telnet connections I would say
that it's a toss-up.  In fact there may be better solutions (the EtherRoute
box perhaps?)

If you like job security and being "the keeper of the knowledge" then 
be the person who maintains/installs CAP (lwserv, pap, aufs, etc)! 

There are a number of other boxes out there now.  The GatorBox is a
one-stop solution.  It may still be a little more expensive, and possibly
a little slower (it does a lot of work).  I'm planning on buying two GB's 
(with the works) for a network installation I'm doing later this year.  
I'm still investigating the market but I haven't seen a box that will 
do everything that the GB does.  
-- 
John Kupec, Agouron Pharmaceuticals, Inc., La Jolla, CA
jwk@scripps.edu or uunet!agouron!kupec

terryr@fogno.corp.sgi.com (Terry Reynolds) (03/26/91)

if you get anything good please share...  We have gators after tossing
Kinetics in favor of the Appleshare stuff on the gator, but now are
considering adding kinetics for the telnet capability.  We've ordered
a kinetics to fool with, but...
-- 







 Tidbit of the day:   Please refrain from running down your fellow
employees.
----------------------------------------------------------------------
-----------
Ms. Terry Reynolds, datacom   |    What if there were no such thing
as
terryr@corp.sgi.com           |           a hypothetical question?
______________________________________________________________________
___________

anders@verity.com (Anders Wallgren) (03/26/91)

What is lacking in the GatorBox that you need to add a kinetics 'for
the telnet capability?"  The GatorBox does dynamic address
assignments, in case you weren't aware of this.

anders

pierre@imag.fr (Pierre LAFORGUE) (04/23/91)

We are using 18 Kinetics (including 1 KFPS1 -the first in France...) and 2
Gatorbox. All are centrally managed by an Unix atalkad daemon.
Our softwares: Eudora, MacIP, etc on the Macintosh, and CAP on Unix systems
(CAP-->appleshare server + remote printing in the both ways), KIP and a
locally upgraded KIP/gw.srec for the old KFPS.

With our experience, I can argue the functionnality of the both are very
close (if you do not buy Gatorshare -even if you prefer NFS/Appleshare
translation instead of CAP, you have better buy a site licence for the NFS
Macintosh client instead of buying Gatorshare).
Maybe the Gatorbox has more future.
With atalkad, it is somewhat easier to configure a KFPS4 than a Gatorbox
-once you understand the meaning of all the flags:-)
Kfps as gatorboxes can talk to each other over IP internets.
Both are able to assign dynamically addresses to the Macintosh.
The support is very very bad for both ... Maybe because we are a non US
academic site.
For instance, this is a mail sent to Cayman without any answer, about a bug:
Date: Fri, 4 Jan 1991 17:06:37 +0100
Subject: Problem with the Gatorbox using IP subnetting

We found a problem in the Gatorbox soft, detected when trying to use
subnetting (one IP subnet on the Ethernet cable, other IP subnet on the
LocalTalk cable). Our solution is rather unelegant, as it implies using
an undocumented feature of the Apple's MacTCP, on every Macintosh ...
Do you know another solution ?  Or maybe a fix in a next release ?

The Tcp/Ip tools running on Macintosh (e.g. Telnet, Ftp, Mail, etc) cannot
dialog with others IP nodes through the Gator gateway: they register a correct
IP address, but they cannot send messages outside the local cable.

The cause is the following:
- The standard "Ip over Appletalk" software uses the following method to
  obtain the address of the Appletalk node where to send IP packets with
  a destination IP address x:
  - The sending Mac sends a NBPQuery request for "y:IPADDRESS@*" (y is the
  dotted decimal representation of the address x). There is no concern of
  local or non-local address, or of net mask.

- With K-star, a KFPS gateway has the following behaviour:
  - If x corresponds to an address which resides on the local LocalTalk
    cable, the corresponding machine (if there is one) sends a response
    with its node address.
  - If x corresponds to an address outside the IP range assigned to the
    local cable, the gateway responds its node address.
  Therefore the gateway declares being the correct AppleTalk node for
  all the traffic leaving the local cable.

- But the GatorBox using subnetting does not answer to any NBP IPADDRESS
  query (except to a query for the gateway address itself).
  Therefore the Macintosh (e.g. MacTcp) does not know where to send its
  outgoing traffic.

We propose a method to circumvent this problem, by the way of a special
configuration of MacTcp on every Macintosh: we set a mask and a gateway
address in the MacTcp control panel dialog. These settings are not allowed
in the dialog window when the option "server" is selected.
In fact we must change the `obtain address' configuration from `server'
to `manually', then set the mask and the gateway, then restore
the `obtain address' configuration to the correct value, i.e. `server'.
Of course such a trick is specific to MacTCP ; we did not found the same
way inside non MacTcp softwares, e.g. the old MacIp 3.x and Telnet NCSA V2.

Yours truly,
Pierre Laforgue
(Director of computer resources, IMAG Institute, Grenoble, France).