It depends too on what you are doing in the lab.
Windows 95 is probably preferrable to Windows NT for simple office-
like applications. It's networking support is a sort of hit-or-miss
situation, particularly if the machine is inter-networked with non-
Windows machines. Windows 95 is semi-stable (it should be noted that
both MS Office and Explorer 4.0 contain bugs that do corrupt some
of the internal structures used by Windows and that can lead to
unexplainable and unrecoverable software failures).
Windows NT is typically useful only as a secure form of Windows 95.
It require more system resources, networking is somewhat less robust,
and there are fewer hardware drivers than for Win95. NT can double
as a server for a small network, but doing so can sometimes preclude
using it as a workstation in its own right. NT is alos not nearly as
stable as you would hope either.
There's also Linux. This is typically a good choice if you have
custom applications or more computer-intensive applications. UNIX
operating systems have a steeper learning curve, but they do give
you a level of computing power not easily had in the Windows world.
Linux can run on cheaper hardware, supports a wider range of hardware
than either Windows, inter-networks nicely with UNIX, Windows, and
MacOS, requires less resources, has less overhead and is far
cheaper (free, as opposed for the $99 for Windows 95 or the $495+
for NT). Linux distributions tend to come with considerable amounts
of development software as well. At least the Red Hat 5.0 distribution
is far easier to install than Windows 95 or NT. It's a very popular
replacement for NT in many environments. However, Linux may require
that one person accept the job of administering the machine (NT
woudl require this too). Now, the administrator need not be
physically be present to administer the machine (it can all be done
over a network), and a competent administrator can configure the
machine from a reference machine, or distribute software updates
with virtually no effort, but you need to find the person to do it.
It is our experience that our customers typically have a
heterogeneous collection of Windows and UNIX machines. Windows is
used primarily for running office while UNIX machines are typically
used for databases, computational apps (we have a lot in
bioinformatics), batch computation, image processing, and various
low-level networking jobs (often as a file server for the PC or
Mac running the office apps). We've since seen a shift from primarily
Solaris towards more Linux (lower cost, can be installed such that
both Windows and Linux are on the same machine). Many of our
customers are also moving to Linux as their primary platform for
reading mail, WWW browsing, and more interactive biocomputing
applications. So, this may be the reference platform for
bioinformatics in the next few years (just in case, our products
are available for NT, Solaris, Irix, DEC, and Linux).
You know your staff best and you may very well want to ask them
which platform they would prefer. (It should also be noted that
you can get WordPerfect, etc. for Linux, as well as ApplixWare
which is a very capable office suite).
Rex A. Hoover wrote in message <6ab500$7ja$1 at winter.news.erols.com>...
>"opinions on using Windows95 vs WindowsNT"
>>>I recommend going with Windows 95.
>We have 2 IBM personal computers in our lab connected via a 10Mbps LAN.
>Both PC's have Windows 95 on them and setting up the LAN was a breeze.
>We have an IBM 300PL with mother board based 10/100Mbps Ethernet and
>an IBM 300GL with a 10/100Mbps Ethernet Network Interface Card
>connected via RJ-45 cables to a 10base-T 8 port concentrator from Boca
>Research Inc. The concentrator, NIC and cables cost a grand total of $200
>we can share files and printers with ease. We also have been using the
>for dial-up access and required nothing extra to make it work.