In article <199510181905.MAA28651 at net.bio.net>, jperry at UWCMAIL.UWC.EDU wrote:
> Can anyone offer an objective, non-prejudicial appraisal of what one
should be
> purchasing in setting up a computer lab for biology?
I'm not exactly non-prejudicial, and I'm not sure about biology
requirements, but I do the mac support for our rather large chemistry
department, and would tend to go with at least mostly macs. They are far,
far, easier to install and maintain. We have roughly equal numbers of
macs and pcs (about 250 each) but the computers that really get used are
the Macs, and if you look at the age distribution, few new pcs are being
purchased compared to new macs. The existing research pcs are primarily
connected to instruments for data collecting, and sometimes analysis,
though it's becoming more common for people to shift their data to either
a mac or a unix box for analysis (depending on user preferences and type
of data.)
We have three computer labs for classroom instructional use: Silicon
Graphics Indigos (primarily to run some specific molecular graphics
software, but they got used for other things like Mathematica too), about
30 286/386-era pcs for some of the undergraduate labs, and a newer
undergraduate lab of 25 PPC 6100s. The mac lab is in nearly constant use
with almost no maintenance on my part, other than making sure that things
are installed properly and doing any upgrades before each quarter. My
guess is that I spend roughly 6 to 10 hours per quarter on minor fixes and
new installs for the lab, with about a week during the summer for clean up
and upgrades, so two weeks per year total. The pc lab (not networked) is
still hanging in there, mostly for data collection, but it has to be
supported by the teaching-lab techs, who spend a fair amount of time on
it; they guessed 3-5 hours/week when I asked once, but they don't keep
close track; at the 3 hour/week rate, that's about 4 weeks per year. It's
unlikely to be upgraded in the near future, but since they do simple
tasks, there's not much need to upgrade until they start to fall apart.
We have essentially 3 computer support people for the department (not
counting grad students who maintain their own machines), with a division
roughly of one full time mac person, one full time unix person and one
full time pc person (we actually overlap somewhat). One full time mac
person is just about able to support 250 macs, mostly on ethernet, of all
varieties and ages, and multiple uses. In addition to maintaining the
hardware, we do some informal software training, especially for the
secretaries. There is a backlog; we can do problem fixes pretty quickly,
but sometime new installs are delayed, and as more macs show up, it's
getting worse :-). The pc person could only really handle about a third
of that load at the level of support we give the macs, so pcs simply lose
out and much of the support by default falls on grad students, with the
result that a lot of the pcs on our records are functionally doorstops
now. Interestingly, the only macs we have that are unused are 3 broken
Mac Pluses and 2 broken SEs which are waiting for someone to have enough
time to merge them into a couple of working ones (already spoken for)...
There was an interesting article in our local newpaper last spring about
Intel. They allow one support person per ~40 pcs, and were horrified to
find a 'renegade' group using ~100 macs with a part time person handling
all of the support :-) This has certainly been my experience, and that of
the other support people that I talk to, admittedly rather biased towards
mac people...
> Or data collection to integrate into the biology laboratory?
There are more options, especially cheap options, for data collection on
pcs. There is stuff available for macs, but inertia and supply and demand
favors pcs. On the other hand, if you just need to connect to a box that
squirts ascii data to a serial port at a moderate rate, there is plenty of
software on the mac available to collect it and either feed it into an
analysis program, or analyse it on the fly (such as Igor, a programable
plotting package). One of our faculty is starting to experiment a bit
with collecting data through the PPC microphone port to use the built-in
analog to digital converters, but I haven't heard lately how it's coming.
> Is the best of both worlds a dual platform Mac (with DOS card)?
We have one that I played with a bit. It isn't a full pc: there is no
parallel port, and if you need to run software protected with a dongle
(unheard of on a mac, but I'm told it's not uncommon on pcs) I would
expect it to not work. This could possibly also be a problem for data
collection hardware, if it goes through the parallel port. If you need
ethernet there is the additional problem that only one processor can be
connected to the net unless you fiddle with the network configuration
files on both the mac side and the pc side (not something we permit our
users to do on their own, as mistakes have the potential to disrupt
network traffic) and then reboot. There's no easy way to have the net
connection just switch when you do because both processors can run in the
background; the ether hardware has to be explicitly handed to one of the
processors at boot. Other than that, the few simple things we tried out
worked fine (Windows 3.0, MS Office, Word Perfect).
> Does anyone have a vision of what will be most useful for CD ROM materials
> (Mac vs DOS) in the future?
Almost all Macs now ship with a quad-speed CD-ROM (essential! It's getting
darned hard to boot from a floppy these days :-)). The Apple built-in
software reads all of the standard formats, and all current macs will pass
audio through to the speakers with no hastle or extra hardware. Most
software that I've looked at has both mac and pc versions, often on the
same disk. But I haven't done an in-depth search. It is interesting to
note, however, that even though macs 'only' have 8% of the hardware
marketshare, they have about 35% of the software market share.
> Are there any reviews that anyone is aware of that addresses this issue?
The trade rags often do this sort of thing, but as you might expect, they
aren't at all unbiased, and never actually compare oranges to oranges
(much more likely to compare peanuts to elephants!). Worse, even when
they do a review of a piece of scientific software, they almost never look
at the right things (how flexible is it at importing 30 year old data?
Does it force you to use double precision at all times? Can the plotting
handle error bars?)) You'll get better advice from talking to other
academic departments around the country.
> We're needing to make some decisions associated with a major building
project.
Be sure to assume that all computers (no matter what type) will be
directly on ethernet. The general rule is to run at least one twisted
pair connection to each room (you can expand that with hubs later, which
are now pretty cheap), maybe several ports for large labs, though again,
you can always add on hubs, and thin-wire goes a long way (but then you
have to plan for thin-wire on the computers, and twisted pair (10BaseT) is
much more common now (and built-in to some of the newer macs). It's
_much_ cheaper to do the initial wiring while there are already holes in
the wall!
> Our present computer inventory (none diedicated to the natural sciences) is
> virtually all DOS-based hardware, but there is a contingent that really
thinks
> Macs are better.
There's no reason not to go with a mixed shop (assuming that the pc users
really want to keep their pcs and don't want to switch to macs). For all
of the improvements in windows 3 and 95, and the surface similarities,
when it comes right down to it, they are still _very_ different from a
mac. Mac people often really truly hate being forced to use intel/ms
stuff, and vice versa, and there's nothing wrong with that (either
direction). I hate brocolli, and a friend of mine hates beef, and there's
nothing wrong with that either :-). People certainly get more useful work
done when they don't spend a lot of time swearing at being forced to use
equipment that they're uncomfortable with. There should be room for
personal preference in the world, especially at a university. A mixed
shop could also have the advantage of exposing students to more than one
paradigm. As for interchange of data and word processing files, that's
not usually much of a problem these days. Most common programs can read
and write each others formats, even between macs and pcs. If you need
equations, the best solution we've found is to use the full version of
Mathtype rather than the subset built in to Word, and then equation
conversion is automatic, and there's no problem with different versions of
Word, or the people who prefer Clarisworks (or Framemaker, Word Perfect,
or whatever). Converting embedded pictures can be messy, but that's
getting better every year, too. Not everything will convert completely
between different programs, such as Clarisworks and Word, but that's
program dependent rather than platform dependent. The current crop of
laser printers (though not quite the very cheapest yet) can usually talk
directly to both macs and pcs (and unix and VMS), either directly over
ethernet (preferred) or over a combination of ether, serial and appletalk,
wth a minimun (or even no) special configuration.
I hope this rambling has been of some use...if you have further questions,
I'm always willing to ramble some more :-) It might help to know in more
detail about what you need: how many computers (25? 100?), who will be
using them (faculty? grads? undergrads?), for what (set lab analysis? a
single lab? multiple labs? research lab analysis? how much data? word
processing? do you intend to standardize on software? will they be
networked?).
--
Allyn Weaks
allyn at u.washington.edu
PNW Native Wildlife Gardening: (under construction)
http://chemwww.chem.washington.edu/natives/
