21 Aug 90 Don Gilbert biocomputing office indiana university A question that I posed on the bionet.software newsgroup recently is: Could everyone who has an opinion on what the most powerful &/or most popular multi-user system for molecular biology will be in the next 5 years, please reply to me? My thanks to those who took the time to answer -- many of whom are in the positions of aquiring, administering and programming for these multi-user computers in molecular biology. The responses are quite useful to me and I hope to others looking to the future of molbio computing. This was not an unbiased survey -- my opinions colored the question and the Bionet newsgroup is widely distributed thru the Usenet of Unix computers. It is also distributed outside of Usenet, however. Of 17 responses, about 15 saw Unix as the current or near future leader for multi-user systems. Most respondents now use a Unix system. Of those using VMS, some are migrating to Unix, and some are using it only to run specific software such as GCG. I asked this question with the implicit assumption that some form of large, multi-user computers would remain a necessity for molecular biologists in the near future. Several respondents corrected or challenged this assumption by discussing mac/pc networks as a replacement for multi-user computers, or at least to use in a majority of data analyses. My opinion remains that large, non-personal computers will continue in the forseable future as essential to molecular biology. They are often now, and will increasingly become compute servers, offering raw computing power to networked users who will use them, but in a microcomputer environment. Examples are the search servers provided by GenBank and EMBL, and one mentioned by Alex Reisner. My plan now is to look for a Sun SparcStation or Silicon Graphics Iris to add a Unix, RISC computer to our local VMS microVax. I also tested the real-world performance of a Sun 4/330 rated at 20 MIPS versus Iubio rated at 0.5 MIPS. Using FelsensteinÕs DNAML on 18 species of 128 base strands of aligned 5S rRNA, IuBio took 33 hours and the SUN took less than 2 hours. IuBio cost us around $55K a couple years back. A 16 MIPS SparcStation 330 will cost us about $25K. The full text of these responses should be quite useful to anyone faced with similar decisions about their future multi-user computing platforms. The complete text can be obtained by anonymous ftp to iubio.bio.indiana.edu, directory [Archive.Help] as ÔUnix-in-MolBio.txtÕ, or ask me for an e-mailing of this. Summary of responses: douglas prasher, woods hole oceanographic inst: has unix, plans for unix, uses IG suite, MASE, Phylip Joe Felsenstein, Dept. of Genetics, Univ. of Washington, Seattle: has DECstation, DEC and Sun are main workstation contenders, Unix with X-windows mouse/window interface is the portable route of the future. stern@CSHLAB.BITNET: prefers unix Roy Smith, Public Health Research Institute, New York, NY: has unix, likes mac networks, discusses pros of mac network, cons of unix administration. Ross Smith, Cell Biology, NYU Medical Center: has vms, plans for vms or vms/unix convergence. Joe Smith, U. of Pennsylvania, Dept. of Biochem. and Biophysics: has unix, plans for unix. Henry Ford (@gdr.bath.ac.uk): has unix, plans for unix, thinks european data bases are planning unix migration. Bob Woodman (@hpuxa.ircc.ohio-state.edu): has unix, likes OS/2 networks. John Cargill, Ontario Cancer Institute, Toronto: has mixed vms and unix, plans for unix; unix is rougher around edges, harder for system management; several helpful comments for vms->unix migration. Paul Schneeman, Computing Services, UCDavis: has DECstations, vms and unix, prefers vms; comments support vms over unix. Una Smith, Princeton Univ.: has unix, supports what works best for software, likes micro networks and Iris computer. Rick Westerman, Biochemistry, Purdue University: has unix and vms, prefers unix or mac/pc network. Sue Funk, Chemical Abstracts Service, Columbus, OH: has unix and vms, prefers unix but uses vms for the GCG software. Dan Davison, dept. of biochem. & biophys. sci., univ. of Houston: prefers unix Mike Cherry, Massachusetts General Hospital, Boston: sees unix dominating in future, but with mac/pc networks as main user front end. Has thoughtful comments on future of molecular biology computing. Alex Reisner, Sydney University: has GCG running on unix Sun SPARC Server now, prefers Sun SPARC for future. Tony Kyne, Walter and Eliza Hall Institute of Medical Research, Victoria, Australia: both vms and unix are needed now, Sun is unix machine of choice, more good comments. >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> From: SMTP%"prasher@hyatt.whoi.edu" 16-AUG-1990 17:18 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Thu, 16 Aug 90 18:11:38 EDT From: prasher@hyatt.whoi.edu (Doug Prasher) Message-Id: <9008162211.AA00166@hyatt.whoi.edu.whoi.edu> To: Don.Gilbert@iubio.bio.indiana.edu Subject: Re: Unix vs. other multiuser computers? Don, We're using Unix os on a bunch of Suns. I personally have very little experience with VMS and I'm certainly not crazy about it. We're running several MB-related programs on our system: IG Suite, MASE, PHYLIP. It's a great system - there's nothing else to say. douglas prasher woods hole oceanographic inst prasher@hyatt.whoi.edu >>>>>>>> From: SMTP%"joe@genetics.washington.edu" 16-AUG-1990 17:38 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? From: Joe Felsenstein Message-Id: <9008162230.AA12349@evolution.genetics.washington.edu> Subject: Re: Unix vs. other multiuser computers? To: Don.Gilbert@iubio.bio.indiana.edu (Don Gilbert) Date: Thu, 16 Aug 90 15:30:51 PDT In-Reply-To: <9008162127.AA03875@genbank.bio.net>; from "Don Gilbert" at Aug 16, 90 9:25 pm X-Mailer: ELM [version 2.2 PL0] Don -- Sun SPARCs will obviously be big because they're letting other outfits license the instruction set and/or use the chip set. So you will see many variants of it from different companies. I am partial to my DECstation 3100 which is really a MIPS chip. I don't know that DEC will take over from Sun but they and Sun are two of the biggest contenders in the workstation market. If you write for Unix with X-windows mouse/window interface you should be able to port the results to any of the major Unix workstation families, so exact manufacturer may not matter. At some point we'll have GCG and IG software for anything that talks Unix and X. So I don't know that you have to choose just one manufacturer, though it is convenient to have all the machines in one lab or department be the same make. Joe ----- Joe Felsenstein, Dept. of Genetics, Univ. of Washington, Seattle, WA 98195 Internet/ARPANet: joe@genetics.washington.edu (IP No. 128.208.128.1) BITNET/EARN: FELSENST@UWALOCKE UUCP: ... uw-beaver!uw-entropy!uw-evolution!joe >>>>>>>> From: SMTP%"stern@CSHLAB.BITNET@CUNYVM.CUNY.EDU" 16-AUG-1990 17:52 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Return-path: stern%CSHLAB.BITNET@CUNYVM.CUNY.EDU Date: Thu, 16 Aug 90 18:45:41 EDT From: stern%CSHLAB.BITNET@CUNYVM.CUNY.EDU Subject: Re: Unix vs. other multiuser computers? To: Don.Gilbert@iubio.bio.indiana.edu Message-Id: <9008162245.AA10046@cshl.cshlab> I vote for UNIX. And I like Silicon Graphics and Sun, but then,.. who doesnt? >>>>>>>> From: SMTP%"roy@alanine.phri.nyu.edu" 16-AUG-1990 19:03 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Thu, 16 Aug 90 20:15:23 EDT From: Roy Smith Message-Id: <9008170015.AA11091@alanine.phri.nyu.edu> To: Don.Gilbert@IUBio.Bio.Indiana.Edu Subject: Re: Unix vs. other multiuser computers? Newsgroups: bionet.software In-Reply-To: <4094@cica.cica.indiana.edu> Organization: Public Health Research Institute, New York City In article <4094@cica.cica.indiana.edu> you write: > Could everyone who has an opinion on what the most powerful &/or most > popular multi-user system for molecular biology will be in the next 5 > years, please reply to me. I am a long-time unix hacker (cut my unix teeth around 1977). I can't concieve of working in any other environment for most of my software development work. Around here, we have a multiuser vax running Unix and a bunch of Sun workstations, also running Unix. We run a variety of sequence analysis software, including the MBIR/Eugene package. The idea of working on a VMS system gives me the shakes. Now, with that out of the way, I guess I should get around to answering your question. I'm not sure about the most powerful, but I'm fairly certain that the most popular system will be the Mac. No, it's not multi-user, but who cares? Get youself a big AppleShare file server, put a lot of Mac workstations, laser printers, netmodems, etc, on the wire, and there you are. No, it's not "a multi user system" by strict definition, but who cares? It's still a bunch of users, sharing files and peripherals, and that's all you really care about, right? I used to poo-pooh macs. I resisted at first. But now I'm in love with them. I have spent years attempting to bring people up to speed on Unix systems. Some catch on, some don't, but it's always a struggle. On the other hand, sit somebody in front of a Mac with some well-designed software, point them in the right direction, and you will probably have that person doing productive work by the end of the day. There are already some sequencing analysis packages available on the Mac. Some are good, some are terrible. Some are free, some are horribly expensive (Note: there is not, as far as I have been able to tell, much correlation between price and quality in software). More and more software is getting written, and it's only a matter of time (in my opinion) before Macs push Unix systems to the wayside, at least for use by non-computer-professionals. The other down-side to Unix is that it is not easy to administer. It takes a professional Unix type to keep a Unix system running properly. I've seen a bunch of biologists set up, almost entirely on their own, a 6 workstation Mac network, complete with file servers and printers, and keep it running well with only very occasional need to ask me for help with some problem. /roy -- Roy Smith, Public Health Research Institute 455 First Avenue, New York, NY 10016 roy@alanine.phri.nyu.edu -OR- {att,cmcl2,rutgers,hombre}!phri!roy "Arcane? Did you say arcane? It wouldn't be Unix if it wasn't arcane!" >>>>>>>> From: SMTP%"roy@alanine.phri.nyu.edu" 16-AUG-1990 20:57 To: GILBERTD Subj: Re: Unix vs. other multiuser computers? Date: Thu, 16 Aug 90 22:09:35 EDT From: Roy Smith Message-Id: <9008170209.AA11237@alanine.phri.nyu.edu> To: GILBERTD@IUBIO.BIO.INDIANA.EDU Subject: Re: Unix vs. other multiuser computers? For a 24-hour rna fold, there are several possibilities. 1) Use a version of rnafold that runs on a native mac and just the the dumb thing sit there for a day and much rna. 2) same as 1 but let it run in the background under multifinder. Problem here, is my experience with background tasks under multifinder is that there isn't much cpu left to do anything in the foreground with. I suppose it all depends on whether you have a Plus, a cx, or an fx. Anybody who does rnafold on a Plus deserves what they get :-). 3) Open up a window to some big numbercruncher off in networkland somewhere and do the rnafold there. The network numbercruncher could run Unix, VMS, or whatever, as long as it crunches fast and you don't have to talk to it much. If it's really ugly, have a nice Mac front end that does the talking for you (like Mathematica does, with a Mac front end and a remote kernel). I absolutely agree, $3k per seat for MacVector is a bad joke. Perhaps, however, I misunderstood your original question. In my experience, the big rna crunching run is only a very small fraction of what biologists use computers for. Yes, there are people who spend their whole lives folding rna, but for most people, restriction maps, GenBank/Dayhoff searches, homology comparisons, hydropothy plots, and the like are what gets done most of the time, and a decent size Mac (say a 4 meg cx) is plenty of power for most of that (well, not the fasta runs, but you ship that off to the numbercruncher in the background, or mail your search to genbank.bio.net, or whatever). Actually, even more common than any of the above is plain old word processing, for which I don't think anything beats a Mac for most purposes. So, in summary, you have two kinds of people. "Just plain users" who will do the majority of their work on a Mac and send the occasional big job to a back-end numbercruncher (probably a big Unix box, but maybe not). And, power-users, who will still want a Mac for word processing, doing figures, slides, etc, and will use it as a front end to the back-end machines, just like the plain user, except more often. Then, of course, are the specialized machines like Irises. Another reason to get a Mac is that software is *so* much cheaper. Take my two machines, a IIcx (5 Meg, 40 Meg disk, 19" mono display) and a Sun 3/50 (8 Meg, diskless, 19" mono). Both 680[23]0's running at comperable clock speeds. Essentially much the same hardware, and both essentially single-user machines. Price a good word processor for the Mac (I like MS Word 4.0). About $250 at MacConnection, about $100 at the NYU bookstore, bundled with SuperPaint, WordFinder, and some macro package I don't use. Now, price a comperable WYSIWYG word processor the the Sun, say Frame. Last I looked, about an order of magnitude more. Of course, there is lots of high-quality free software for both, so maybe that's a bit of a straw-man argument. Looking forward to seeing what other people's opinions are. /roy >>>>>>>> From: SMTP%"roy@alanine.phri.nyu.edu" 16-AUG-1990 21:02 To: GILBERTD Subj: Re: Unix vs. other multiuser computers? Date: Thu, 16 Aug 90 22:14:26 EDT From: Roy Smith Message-Id: <9008170214.AA11245@alanine.phri.nyu.edu> To: GILBERTD@IUBIO.BIO.INDIANA.EDU Subject: Re: Unix vs. other multiuser computers? PS, I get bad vibes about A/UX. If you really want to run Unix, I can think of better platforms to run it on (Spark SLC comes to mind as the latest hot Unix box). You buy a Mac (in my opinion) because you like the Mac interface. If you want to run Unix, the Mac doesn't seem like a particularly cost-effective platform to run it on. Granted, I have never used A/UX, so I don't know how well it integrates the Unix and Mac Finder environments. On the other hand, Running something like NCSA Telnet under Multifinder gives me a pretty good integration of the two, so I havn't seen much need to experiment with A/UX. Of course, if you can't afford both machines, A/UX may be the only way to get both environments in one box. /roy >>>>>>>> From: SMTP%"SMITH@MCCLB0.MED.NYU.EDU@CUNYVM.CUNY.EDU" 16-AUG-1990 19:59 To: DON.GILBERT Subj: Unix vs. other multiuser computers? Date: Thu, 16 Aug 90 20:55 EDT From: "Ross Smith: (212) 340-5356" Subject: Unix vs. other multiuser computers? To: don.gilbert@iubio.bio.indiana.edu Message-id: <4B3D6FFC0ADF001966@MCCLB0.MED.NYU.EDU> X-Organization: NYU Medical Center, 550 1st Ave., New York, NY 10016 X-Envelope-to: don.gilbert@iubio.bio.indiana.edu X-VMS-To: in%"don.gilbert@iubio.bio.indiana.edu" X-News: mcclb0 bionet.software:269 >VMS has an air of obsolescence about it -- today's >programmers write for Unix first, and if time permits, someone ports it to >VMS. Well we write for VMS first! >I won't touch the VMS file handling system, whereas Unix is duck >soup. The good sequence analysis software from GCG is the only thing >really tying us to VMS, and it sounds like a Unix version is near. VAM/UNIX convergence is just fine, everyone will benefit when it gets here. From my point of view, however, the relative simplicity and power of VMS makes it an attractive system for now. And our 7 MIP VAX still outperforms the GenBank FASTA server (80 MIP Solbourne), so why move? +---------------------------------------------------------------------------+ |Ross Smith, Cell Biology, NYU Medical Center, 550 First Ave., NYC, 10016| |Phone: (212) 340-5356: FAX: (212) 340-8139 (Alternate NYUMC) (212) 340-7190| |E-Mail: SMITH@NYUMED.BITNET (BITNET), SMITH@MCCLB0.MED.NYU.EDU (Internet)| +---------------------------------------------------------------------------+ >>>>>>>> From: SMTP%"jes@mbio.med.upenn.edu" 16-AUG-1990 22:32 To: DON.GILBERT Subj: Unix vs. other multiuser computers? Date: Thu, 16 Aug 90 23:25:54 EDT From: jes@mbio.med.upenn.edu (Joe Smith) Posted-Date: Thu, 16 Aug 90 23:25:54 EDT Message-Id: <9008170325.AA03595@mbio.med.upenn.edu> To: Don.Gilbert@IUBio.Bio.Indiana.Edu (Don Gilbert) In-Reply-To: Don.Gilbert@IUBio.Bio.Indiana.Edu's message of 16 Aug 90 21:25:29 GMT Subject: Unix vs. other multiuser computers? As long as you're *asking* for people's 'religous' persuasion... Unix is the only computing environment I will spend any amount of time on - either using or programming. I'm convinced that a real multiuser, multitasking environment is the most productive one (even if there is only one person using it!). People in the real world do not have time to know more than one computing environment, and the only one with any future is Unix. I don't want to change horses ever again, and I'm convinced Unix is the only realistic (if somewhat ugly) choice. Unfortunately, many of the nice features of PCs & Macs are lacking in Unix, but it has lots of room to improve, without losing it's power. The basic guts are there, it just needs to be tuned. Witness the advent of standard, powerful GUIs and desktop managers. I'm betting that within the next 2 years, you'll be able to buy a workstation-class Unix box that's as canned as a Mac for the price of a nice PC today (3-4K$). As far as molecular biology, there is a lot of inertia in VMS, PC and Mac environments. There are still a lot of natural scientists for whom 'computer == VAX'. But that's changing as well because the software developers, as you pointed out, are often going to Unix first. We just evaluated two sequence gel readers. One ran on a (generic) Sparcstation, the other on a PC with a camera and a fancy, proprietary, graphics subsystem. The price of the hardware was essentially the same in the two systems - which would you choose? This is the kind of choice that will jolt the typical molecular biologist into the Unix wagon. I have only had experience with our lone Sun-3, so I won't venture an opinion, except to say that I'm very pleased with it's performance. >>>>>>> From: SMTP%"H.Ford@gdr.bath.ac.uk" 17-AUG-1990 03:44 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Fri, 17 Aug 90 9:30:31 BST From: Henry Ford To: Don Gilbert Subject: Re: Unix vs. other multiuser computers? 1: I believe that the European data bases are planning on moving over to unix machines in the next three or four years. 2: Unix is not necessarily difficult to teach - biology students here pick it up quite quickly, and are soon into shell programming (three weeks or so). 3: Our biochemists use IRIS and love them. We have a couple of sparc machines (not SUNs) and the development people say fine for single users - no good for 6 compiling programs! They haven't tracked the problems down yet. 4: I'm a ex-biologist in a computer unit dealing with a wide range of biological software - mainly statistical. Henry Ford >>>>>>>> From: SMTP%"drbob@hpuxa.ircc.ohio-state.edu" 17-AUG-1990 08:16 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Fri, 17 Aug 90 09:10:43 edt From: Robert H. Woodman To: Don.Gilbert@IUBio.Bio.Indiana.Edu Subject: Re: Unix vs. other multiuser computers? Newsgroups: bionet.software In-Reply-To: <4094@cica.cica.indiana.edu> Organization: Ohio State Univ IRCC In article <4094@cica.cica.indiana.edu> you write: >I'd like to hear from others in molecular biology about your favorite >flavor of multi-user computer operating system. Don, I'm not a molecular biology software developer, just an end user. Right now, Unix looks to be the most powerful system, and it may remain so, IMHO, for the next 5 years. However, if Microsoft gets its act to- gether wrt OS/2, I believe that in the next 10 years OS/2 will be the system to compete against. After all, most convenient DOS programs will port up to OS/2, and most Unix programs can now port over to OS/2. As PC's get more powerful, I think that OS/2 will be the winner *if* Microsoft can get it moving. Bob Woodman woodman.1@osu.edu _OR_ drbob@hpuxa.ircc.ohio-state.edu Disclaimer: Being a graduate student, my opinions are usually ignored. >>>>>>>> From: SMTP%"SYSJOHN@UTOROCI@ugw.utcs.utoronto.ca" 17-AUG-1990 09:07 To: DON.GILBERT Subj: multi-users operating systems/hardware Date: Fri, 17 Aug 90 10:52:00 EDT From: Subject: multi-users operating systems/hardware To: don.gilbert@iubio.bio.indiana.edu X-Original-To: don.gilbert@iubio.bio.indiana.edu, SYSJOHN Message-Id: <90Aug17.095344edt.57686@ugw.utcs.utoronto.ca> Hi, I'm replying to your posting about post-VMS migration. We run a 10 node VAXcluster here as well as SUN UNIX workstations. We are also looking at migration to UNIX over the next few years. We have found UNIX to be *much* rougher around the edges than VMS and this is cooling our enthusiasm for abandoning VMS. The system management utilities of UNIX, in particular, are more primitive and poorly documented. As a result we forsee a larger management overhead in running these systems. Our strategy is to bring in unix systems for specific tasks. One system is used for logic programming and statistics. Another is used for image analysis. A new system (IBM RS6000) will be used for molecular modeling. When we have a critical mass of UNIX applications (and users) we will look at moving the bulk of our work to UNIX platforms. Fortunately for us the DEC CLSG (software deals) program and DEC WORKSTATION program (hardware deals) keeps the cost of running VMS competitive - even though UNIX RISC systems offer a 2-3 times increase in basic hardware performance. There is a persistent rumour that DEC may port VMS to RISC hardware soon. This move, combined with the rise of the POSIX standard programming interface to minicomputer operating systems, may breathe new life into VMS. If not, I fear VMS will increasingly be an operating system for business rather than research sites. ------------------------------------------------------------------------------- John Cargill | Bitnet: CARGILL@UTOROCI Assistant System Manager | Phone: (416) 924-0671 ext. 5306 Ontario Cancer Institute | 500 Sherbourne Street | Toronto Ontario Canada | M4X 1K9 | ------------------------------------------------------------------------------- >>>>>>>> From: SMTP%"prschneeman@ucdavis.edu" 17-AUG-1990 12:36 To: DON.GILBERT Subj: RE: Unix vs. other multiuser computers? Date: 17 Aug 90 10:26:13 PST Message-Id: <9008171026-05699@MIZAR.UCDAVIS.EDU> From: prschneeman@ucdavis.edu (Paul Schneeman) To: Don.Gilbert@iubio.bio.indiana.edu Subject: RE: Unix vs. other multiuser computers? Don-- We are adding 7 DECstation 5000/200 risc workstations to a lab dubbed "scientific visualization workstation laboratory" and await delivery. I have a strong VMS viewpoint, but I have given up trying to convince the Unix-biggots that perhaps their viewpoint is outmoded. The reason for the decstations include (1)much higher resolution color than Sun could offer, (2)the non-availability of an IBM 6000 offering (troubles with the operating system, according to my Unix sources) (3) an ability of the Decstations to work fairly seamlessly with our VMS systems e.g. the user interface allows switching the tilde between tilde and [escape] in differnt windows. The Motif GUI also convinced me that as an applications person, I no longer really care what the operating system is below the interface. GCG told me that the decstations were the highest priority port for them, and my examination of the MIPS fortran (which is what the DECstations use) is robust enough to do it. One parting shot--Unix people love to complain about the poor development environmnt of VMS. However, you will rarely find one who has actually looked at the compiler quality for large application programs, and those who do then write for VMS. As the numerical recipes in C people wrote-- scientists are people who solve todays problems on yesterdays computers, while computer scientists seem to have the problem reversed! Good luck Paul Schneeman, Computing Services UCDavis The usual disclaimers apply >>>>>>>> From: SMTP%"UNASMITH@pucc.PRINCETON.EDU" 17-AUG-1990 13:54 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Fri, 17 Aug 90 14:16:04 EDT From: Una Smith Subject: Re: Unix vs. other multiuser computers? To: Don.Gilbert@IUBio.Bio.Indiana.Edu >I'd like to hear from others in molecular biology about your favorite >flavor of multi-user computer operating system. We are using a VMS Vax >for GCG software, but I am looking rather longingly for Unix computers >in the future. VMS has an air of obsolescence about it -- today's VMS has always been popular among administrators, not users. It certainly does not conform well to the uses typical academics want to make of it. Typical academics don't want to do things that have been done before, in exactly the same way, over and over again. The UNIX environment is a much more powerful tool, simply because it is so flexible (almost fluid). >programmers write for Unix first, and if time permits, someone ports it to Who wouldn't? ;-) >VMS. I won't touch the VMS file handling system, whereas Unix is duck >soup. The good sequence analysis software from GCG is the only thing >really tying us to VMS, and it sounds like a Unix version is near. Be careful here! There are few mistakes more disastrous than abandoning a platform that supports softwear you *must have* in favor of a better platform that does not yet run this software. No matter how superior the new hardware platform may be, if it doesn't run the codes you need to use, it is not much better than a paperweight. I have seen this happen, and the results are not a pretty sight. >Could everyone who has an opinion on what the most powerful &/or most >popular multi-user system for molecular biology will be in the next 5 >years, please reply to me. I will summarize the responses with a followup >message. Let me also know if you are a molecular biology software >developer. Talk to Greg Hamm, HAMM@BIOVAX.BITNET, at Rutgers University. >Also, if you have a favorite multi-user/workstation computer (e.g., SUN >SParcstation, NeXT cube, Silicon Graphics IRIS, Vax, etc.), I'd like to >hear what it is. First off, I think you should consider very carefully whether you really want a multi-user computer or a network of single-user computers (workstations). If your users would not use graphics on a *daily* basis, you could easily get by with PCs or MACs attached to an ethernet on which a big multiuser system sits. If they use graphics heavily, but are entirely content with MAC or PC capabilities, then you still don't *need* individual workstations, but you might want them. The buzzword now is distributed computing. This means having both big computers and dinky micros, general purpose and highly specialized computers all linked together via one or more network architectures. I got a Silicon Graphics IRIS 4D/220GTX for our lab. It cost less than a DEC uVAX bought at about the same time, and users of both have told me they far prefer the IRIS. The IRIS is also many times faster (I don't have the exact figures at hand, but it's very substantial). While most users continue to think that they must be at the console to use the IRIS, the IRIS is on the campus ethernet, so anyone with network access (ie anyone with a terminal or PC) can get to its LOGIN. Macintoshes are on an Appletalk network (inc. printers), and that network is linked to the campus ethernet backbone. I have 2 hardwired dial-up ports dedicated to PCs, and another PC is hung directly off the ethernet (for incredibly fast file transfers). We're using NCSA Telnet, which is free and very good. We can transfer files between any two computers that we use, more or less transparently. Networking costs were supported by the University directly, and were in any case a very small fraction of the total cost. Here at Princeton the prefered machines are made by SUN and SGI. >>>>>>>> From: SMTP%"cjv@mace.cc.purdue.edu" 17-AUG-1990 09:36 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Fri, 17 Aug 90 09:31:09 -0500 From: cjv@mace.cc.purdue.edu (westerman) Message-Id: <9008171431.AA01635@mace.cc.purdue.edu> To: Don.Gilbert@IUBio.Bio.Indiana.Edu Subject: Re: Unix vs. other multiuser computers? Newsgroups: bionet.software In-Reply-To: <4094@cica.cica.indiana.edu> Organization: Purdue University Cc: In article <4094@cica.cica.indiana.edu> you write: >I'd like to hear from others in molecular biology about your favorite >flavor of multi-user computer operating system. UNIX >We are using a VMS Vax >for GCG software, but I am looking rather longingly for Unix computers in >the future. Same here. We run VAX/VMS because of the GCG software. >soup. The good sequence analysis software from GCG is the only thing >really tying us to VMS, and it sounds like a Unix version is near. I'll believe it when I see it. Porting is always hard and chancy work. > >Could everyone who has an opinion on what the most powerful &/or most >popular multi-user system for molecular biology will be in the next 5 >years, please reply to me. The most popular and powerful system will be the one which has the most popular and powerful software written for it; the hardware should follow the software, not vice-versa. However, if I was to guess at what type of hardware will be popular: Networked MACs (or PCs). Think about it: What do we need the big multi-user machines for? Databases and structure prediction. The other 95% of sequence analysis work can be accomplished more easily on MACs/PCs. A group of networked MACs/PCs will have centralized databases; the structure prediction problems may be solvable in a short enough time with the next generation MACs/PCs, if not, then the problems can always be "shipped out" to mainframes. To prepare for this, I think it is important to get the physical cabling in position. One lab, many lines. The central server could be a UNIX or VMS based machine, but overall it will play inconspicious role in the network. BTW: I think that this is one reason VMS is not as dead as we might think. PCSA and the MAC networking product (I forget it's name) make VMS a powerful central server. >Let me also know if you are a molecular biology software >developer. Not really. But I do support a lot of biologists around here. -- Rick -- Rick Westerman AIDS Center Laboratory for Computational Internet: westerm@aclcb.purdue.edu Biochemistry, Biochemistry building, (317) 494-0505 Purdue University, W. Lafayette, IN 47907 >>>>>>>> From: SMTP%"saf27@CAS@pucc.PRINCETON.EDU" 17-AUG-1990 15:38 To: DON.GILBERT Subj: GCG and UNIX From: saf27%CAS@pucc.PRINCETON.EDU Date: Fri, 17 Aug 1990 16:26 EDT Subject: GCG and UNIX To: don.gilbert@iubio.bio.indiana.edu Don - I saw your posting on bionet.software asking about operating systems. We are also using the GCG package under VMS, and are anxiously waiting for a UNIX version since we are heavily UNIX-oriented and have almost no VMS expertise in-house. I noticed you said "it sounds like a Unix version is near" - have you heard any dates? GCG has been very vague whenever I ask about the timeframe. If/when they do release a UNIX version we will be running it on SUN SPARCstations which we already have. However, I doubt that we can offer you any concrete advice as our interest is in the areas of sequence input and similarity searching in support of our publications, not research. Sue Funk (saf27 @ cas.bitnet) Senior Engineer - Information Systems Chemical Abstracts Service, Columbus, OH >>>>>>>> From: SMTP%"davison@uhnix2.uh.edu" 17-AUG-1990 19:49 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? Date: Fri, 17 Aug 90 19:42:32 CDT From: Dan Davison Subject: Re: Unix vs. other multiuser computers? To: Don.Gilbert@iubio.bio.indiana.edu Cc: software@cica.cica.indiana.edu Message-id: <4A7E5E82E829008CC5@post-office.uh.edu> In-Reply-To: <9008162127.AA03875@genbank.bio.net>; from "Don Gilbert" at Aug 16, 90 9:25 pm X-Mailer: ELM [version 2.2 PL14] X-Envelope-to: Don.Gilbert@iubio.bio.indiana.edu > I'd like to hear from others in molecular biology about your favorite > flavor of multi-user computer operating system. I use Suns, MIPS, Crays (CTSS/COS/UNICOS), Vaxen(VMS/Unix), PCs, Macs, the NeXT, whatever. Use the tool which is best for the job you want to accomplish. "best" can be either (1) you know how to use it best or (2) it is the best tool for the job. > Also, if you have a favorite multi-user/workstation computer (e.g., SUN > SParcstation, NeXT cube, Silicon Graphics IRIS, Vax, etc.), I'd like to > hear what it is. 1) A Cray running Unicos with the FCL shell and X windows via a front end: 2) decent Unix box with X or Sunview dan -- dr. dan davison/dept. of biochemical and biophysical sciences/univ. of Houston/4800 Calhoun/Houston,TX 77054-5500/davison@uh.edu/DAVISON@UHOU "Without the voice of reason, every faith is its own curse" -- Sting Disclaimer: As always, I speak only for myself, and, usually, only to myself. >>>>>>>> From: SMTP%"Cherry@Coffee.MGH.Harvard.EDU" 18-AUG-1990 00:22 To: DON.GILBERT Subj: Re: Unix vs. other multiuser computers? From: Cherry@Coffee.MGH.Harvard.EDU (J. Michael Cherry) To: Don.Gilbert@IUBio.Bio.Indiana.Edu (Don Gilbert) Cc: Cherry@Coffee.MGH.Harvard.EDU Subject: Re: Unix vs. other multiuser computers? Date: Sat, 18 Aug 1990 01:15:26 EDT References: <4094@cica.cica.indiana.edu> Message-Id: <5359008180115Cherry@Coffee.MGH.Harvard.EDU> Organization: Molecular Biology, Mass. General Hospital My opinion: Unix is clearly the future winner between Unix and VMS. An academic can currently buy a 10 MIP workstation for $6000, Sun SPARCStation IPC. The better software is currently available only for VMS systems however this will be reversed in the very near future. The National Center of Biotechnology Information (NCBI) is committed to developing for Suns (Unix), Macintosh and DOS/Windows. They are not developing for VMS! Once Suns can run the GCG package I believe Suns will become generally seen as the better system. The Suns are very cheap and fast with good networking. X Windows or Motif will continue to grow in the engineering environments and thus make some in roads into molecular biology. Another popular computer is the Silicon Graphics Inc. line of graphic workstations. SGI also makes some very fast multiprocessor systems, ie. 180 MIP 8 processor system for ~$250K. The best molecular modelling software is available only for Silicon Graphic systems, there is not very much modelling software available for Suns. The IBM RT6000 is becoming supported by modelling software vendors because IBM licensed SGI's graphics engines and software. Very few of the modelling companies seem to do much business with VMS systems. I don't think VMS will disappear. Digital has stated they are working on RISC VAX processors to be able to compete with SPARCs and other RISC workstations. VAXen will probably become more competitively priced. I see Macintoshes and DOS/Windows machines continuing to grow in number over larger systems. Soon a researcher will be able to use network servers from the NCBI to do all their database searches. Currently the NCBI's database searching software called BLAST is in early tests. BLAST running on a fast SGI server can search all of GenBank and calculate the statisical significance using Karlin's formula in around 20 seconds of CPU time. This will give the molecular biology community near realtime searches from any Internet networked machine. Once the applications and servers start to become available from NCBI others will start using the NCBI toolkit for development. This will increase the portability of the programs and through the use of their subroutine library become a more object oriented process. The ability to interact (in the complete sense of the word) with the database and all the associated reference information removes one of the major justifications for a large multiuser system. File servers will be needed to handle the demand for file storage for all the networked PCs. Well thats probably too much opinion, Mike Cherry Director of Computing in Molecular Biology Massachusetts General Hospital Boston Department of Molecular Biology Massachusetts General Hospital, Boston >>>>>>>> From: SMTP%"reisner@ee.su.oz.au" 18-AUG-1990 20:11 To: DON.GILBERT Subj: multiuser/multitasking OS Date: Sun, 19 Aug 90 11:04:44 EST From: reisner@ee.su.oz.au (Alex Reisner) Message-Id: <9008190104.AA01735@ee.su.oz.au.AU> To: don.Gilbert@iubio.bio.indiana.edu Subject: multiuser/multitasking OS Don, Some short comments. ======================== >I'd like to hear from others in molecular biology about your favorite >flavor of multi-user computer operating system. We are using a VMS Vax >for GCG software, but I am looking rather longingly for Unix computers in >the future. VMS has an air of obsolescence about it -- today's >programmers write for Unix first, and if time permits, someone ports it to >VMS. I won't touch the VMS file handling system, whereas Unix is duck >soup. The good sequence analysis software from GCG is the only thing >really tying us to VMS, and it sounds like a Unix version is near. =================================== With some additional work by us we are running the Sun port of GCG on our SPAC Server 470. We've restricted the graphics drivers to Tek 4014 and PostScript (the Sunport is for GKS) and have a put a shell around the system so that conflicts with environmental variables don't occur. My understanding is that John Deveraux will have V7.0 of the GCG package running under SunOS. While the Sun port isn't quite complete it's pretty good. ==================================== >Could everyone who has an opinion on what the most powerful &/or most >popular multi-user system for molecular biology will be in the next 5 >years, please reply to me. I will summarize the responses with a followup >message. Let me also know if you are a molecular biology software >developer. ========================================= In July we put on line SUSAI (the Sydney University Sequence Analysis Interface running on the SPARC Server. It is accessed Oz wide through AARNet (or equivalent of EARN) and has all the common databases as well as a number of application packages available. The Internet connection we find invaluble for quick downloading. So for example we have MBCRR's plsearch running and have had for several weeks. Currently our software 'development' is in porting VMS progs over to SunOS and database handling software as human resources are rather limited at present. ========================================== >Also, if you have a favorite multi-user/workstation computer (e.g., SUN >SParcstation, NeXT cube, Silicon Graphics IRIS, Vax, etc.), I'd like to >hear what it is. ========================================== At the momnet we're doing our work on Suns. We see that SPARC architecture ought to be around for a fair while particularly with AT&Ts share in Sun on the one hand and the Solbourne/NEC connection on the other. The imminent advent of the Solbourne 5/900 series with the potential of an eight CPM 440 MIPS 96 MFlops system says a lot. Cheers, Alex >>>>>>>> From: SMTP%"TONY@wehi.dn.mu.oz.au" 20-AUG-1990 22:56 To: DON.GILBERT Subj: Unix vs. other multiuser computers? Date: Tue, 21 Aug 90 13:52 +1000 From: "Tony Kyne, Walter and Eliza Hall Institute" Subject: Unix vs. other multiuser computers? To: don.gilbert@iubio.bio.indiana.edu Message-id: <478AA1EAC6FF006F6B@muwayb.ucs.unimelb.edu.au> X-Envelope-to: don.gilbert@iubio.bio.indiana.edu X-VMS-To: IN::"don.gilbert@iubio.bio.indiana.edu" Dear Don, We here at WEHI have decided that we are in an in between state for the present. We have decided to have both VMS systems and UNIX (really SUN OS specifically) to handle our molecular biology needs. It seems there is a wealth of VAX stuff around and a wealth of UNIX stuff emerging. We do not want to deny our biologists any tools because of some blinkered religious belief in computing operating systems. The strength of the UNIX RISC boxes at the moment is their speed for searches etc. We have also decided that there is little point converting code from system to system on a large scale. Our time is better spent filling in the holes that are not covered anywhere. Also, I have a slight problem with the use of the term UNIX. When it comes to dealing with companies for layered software (this more applies outside molecular biology at present where sources are not generally available), you are talking SUN OS, AIX, AUX, Ultrix, RISC Unix, UNICOS, etc etc - they are not binary compatible. The UNIX market is as fragmented as the proprietory systems market if not worse. So I cannot see Becton-Dickenson or Molecular Dynamics or Applied Biosystems or whoever.. supporting all these platforms under the name of UNIX. They could not afford the hardware investment and support. My feeling is that they will support SUN OS. So the question is really VMS or SUN OS. My answer is to have both for the moment. And I suspect from the rumours, press etc, that say by end of 1991 or some time in 1992, I suspect to see a RISC VMS or a combined Unix/VMS system on the one platform or both - you will just switch shells or the like. And I suspect the same will apply for one of the IBM op systems as well. IBM and DEC are not going to toss the towel in and go totally UNIX. WE also see the need for large systems with oodles of memory and lots of grunt. Hence with a group of other institutions we are getting involved with a CONVEX. We want to do larger complete database alignments along the lines of full vectorized Needleman-Wunsch-Sellers with some sophisticated enhancements to improve our searching capabilities. We have found FASTA wanting a number of times especially with conserved proteins with little complete identity. The other alternative would be a vastly parallel system such as a Connection Machine but supply lines to them are not good in this part of the world. So in summary, its VMS and SUN for now to get access to all tools and a bit of speed on the UNIX front, and a CONVEX for some real grunt. Regards Tony Kyne ================================================================================ Dr. Tony Kyne, Head, Computer Sciences Unit, The Walter and Eliza Hall Institute of Medical Research, P.O. Royal Melbourne Hospital, Victoria, 3050, Australia. Phone: International +61-3-345-2586 FAX: International +61-3-347-0852 National 03-345-2586 National 03-347-0852 Email: ACSnet: tony@wehi.dn.mu.oz.au UUCP: uunet!munnari!wehi.dn.mu.oz.au!tony Internet: tony%wehi.dn.mu.oz.au@uunet.uu.net PSIMAIL: PSI%0505233430002::tony ===============================================================================