jim@cs.vu.nl (Jim van Keulen) (01/20/87)
I would like to know more about the Apollo Series 3000 workstations.
These might be an alternative for Sun workstations, but I am not
familiar with the Apollo products. The penetration of Apollo's in
academic circles seems to be much lower then Sun's. Is there a good
reason for this fact?
My principal questions are:
- What is the quality of the UNIX implementation? Is it real UNIX or
just a UNIX shell around a proprietary operating kernel?
- What about the network facilities? Is there a fully transparent
distributed file system?
- How easy is it to port UNIX program distributions to the Apollo?
- What is the general politics of Apollo with respect to make
available kernel sources?
If you compare Apollo series 3000 with Sun-52/M workstations are
there certain advantages of the Apollo over the Sun?
I would be grateful for any information you can give,
Jim van Keulenmark@markshome.UUCP.UUCP (01/21/87)
In article <1036@botter.cs.vu.nl> jim@cs.vu.nl (Jim van Keulen) writes: >I would like to know more about the Apollo Series 3000 workstations. > >- What is the quality of the UNIX implementation? Is it real UNIX or > just a UNIX shell around a proprietary operating kernel? The latter. >- What about the network facilities? Is there a fully transparent > distributed file system? Yes. >- How easy is it to port UNIX program distributions to the Apollo? Depends on what they do. Some things will port. O.S. dependent things that require certain kernel features sometimes will not. >- What is the general politics of Apollo with respect to make > available kernel sources? My understanding is that these are absolutely unavailable. >If you compare Apollo series 3000 with Sun-52/M workstations are >there certain advantages of the Apollo over the Sun? My understanding is that these days (and perhaps always) Apollo is much more aggressive in negotiating special deals for price. My background: I have a lot of first hand knowledge about Suns, have a lot of two year old knowlege of Apollos back when we were making the Sun vs. Apollo decision here at Maryland, and have freshly updated (but second hand) Apollo knowledge based on trying to talk a friend out of buying 200 Apollos. I lost the argument. -mark Spoken: Mark Weiser ARPA: mark@mimsy.umd.edu Phone: +1-301-454-7817 CSNet: mark@mimsy UUCP: {seismo,allegra}!mimsy!mark USPS: Computer Science Dept., University of Maryland, College Park, MD 20742
paul@UMIX.CC.UMICH.EDU ('da Kingfish) (01/23/87)
(I saw Mark Weiser's response to this, can't disagree too much with what he said, but thought I would add this:) In article <1036@botter.cs.vu.nl> jim@cs.vu.nl (Jim van Keulen) writes: >I would like to know more about the Apollo Series 3000 workstations. >These might be an alternative for Sun workstations, but I am not >familiar with the Apollo products. The penetration of Apollo's in >academic circles seems to be much lower then Sun's. Is there a good >reason for this fact? I think that among sites that picked a workstation two or more years ago, most went with Sun, especially if they were interested in Unix. Since then Apollo has come on much stronger in the Unix area, and their Unix is much better (and improving). > >My principal questions are: > >- What is the quality of the UNIX implementation? Is it real UNIX or > just a UNIX shell around a proprietary operating kernel? They have their own proprietary kernel that supports several things that the current bsd kernel doesn't. Their Unix environment ("shell" if you will) runs along with their own interface (aegis). Overall, I think the quality is acceptable. As far as it being "real" Unix, you will quickly find some differences, some things lacking, but none of these has been critical for us. >- What about the network facilities? Is there a fully transparent > distributed file system? Yes, as Mark said. They have a pretty good tcp implementation as well. >- How easy is it to port UNIX program distributions to the Apollo? As Mark said, depends on what it is. I've done uucp, sendmail, and other programs. The kernel interface is certainly different, (there is no /dev/kmem, for example). >- What is the general politics of Apollo with respect to make > available kernel sources? You pretty much can't get them. > >If you compare Apollo series 3000 with Sun-52/M workstations are >there certain advantages of the Apollo over the Sun? > In my opinion, the design of the basic Apollo OS is somewhat more modern than Unix. I chose Apollo over Sun for the machine I use, and University of Michigan has going on 200 Apollos, with more on the way. Even though I like Apollos, I must say that many people with Vax/pdp Unix experience, who like to work in that style kernel, know how to do those device drivers, etc. find the Apollo to be quite different, and don't care for it. I had that opinion when I first started using Apollos, but have developed more catholic tastes in the meantime I suppose. Even here with the multitudes of Apollos, I am quite thankful sometimes for the old reliable Vax 750 with all the 4.3 source code on it. To sum up, I would say that the decision between Sun and Apollo for Unix users is probably not as easy as it was > 2 years ago, and Apollo is worth a look as it stands today. --paul paul@umix.cc.umich.edu seismo!umix!paul
elliott@UTAH-CS.ARPA (Ian A. Elliott) (01/23/87)
> I would like to know more about the Apollo Series 3000 > workstations. These might be an alternative for Sun workstations, > but I am not familiar with the Apollo products. The penetration of > Apollo's in academic circles seems to be much lower then Sun's. Is > there a good reason for this fact? My feelings on the subject are: 1) many academic circles are filled with Unix Wizards, and since the Sun is a multi-terminal Unix environment, these people feel comfortable with the Sun; 2) Apollo has seemed to me to have been thought of more as an engineering workstation, and has not been used to its full extent by many people. My research group (BTW these opinions are my own, and not the University of Utah's which has several Unix guru's and Sun's) has found the Apollo's useful for all sorts of things, which the Sun could only dream of doing. Apollo themselves hasn't even thought of all the things which we have our workstations doing. > > My principal questions are: > > - What is the quality of the UNIX implementation? Is it real UNIX or > just a UNIX shell around a proprietary operating kernel? There is some of the Unix kernel built on top of and along side with their own kernel. For example signals work the same as they do on a VAX. It is not just a Unix shell build on top of AEGIS(TM). > - What about the network facilities? Is there a fully transparent > distributed file system? Apollo has an excellent distributed file system. If you want to get a file off a workstation (node) named "this_node" you simply add "//this_node" in front of the pathname of the file. You can also have soft links which can point all over the place. For example you can have a soft link named "foo" which points to "//this_node/dir1/dir2/foo". > - How easy is it to port UNIX program distributions to the Apollo? This depends on how deep your code goes into the Unix kernel, and whether your code is written according to correct C syntax, or was written to take advantage of holes in pcc. We have seen a few examples of our VAX letting things by that the Apollo didn't. > - What is the general politics of Apollo with respect to make > available kernel sources? My understanding is that they generally don't let too many people at them, but I know that they are available. Flame time: To those who want to live in an exculisively Unix world and never know that anything better exists, then I feel that the Sun is their machine. But for those who are willing to learn a few more things, the Apollo is great. The Apollo's provide more features, better security, and yes, even System V, 4.1 BSD and 4.2 BSD Unix. I have found that developing code on Apollo's is easier than any other computer I have ever tried using (haven't tried the new fancy AI workstations). I am currently porting a tasking package (partly written in 68K assembly and Pascal, to C and VAX assembly) from Apollo's to a VAX running 4.3 BSD, and find that Unix doesn't even come near the development environment which is provided by Apollo. What takes me a few minutes to debug on an Apollo, takes hours or days on Unix boxes. A former member of our group who ported a window package to the Sun said similar things. I heard somebody complain that the Apollo's C compiler wasn't the same as pcc - GREAT! When I can't figure out what the Unix C compiler is complaining about (with such great diagnostics as "syntax error") I use the Apollo C compiler to figure out what is wrong with my C file. Ian ARPA: elliott@cs.utah.edu Snail: Dept. of Computer Science, University of Utah, Salt Lake City, UT 84112
olson@TCGOULD.TN.CORNELL.EDU.UUCP (01/25/87)
I found peoples comments on the Apollo very interesting. But one
topic was not mentioned that I'm curious about. How well does the
Apollo (or Sun for that matter) work in a heterogeneous environment.
I ask because smaller non CS groups (like physics dept.s) don't have
the knowledge to make the machines sit up and beg. Would such groups
effectively be trapped into a single vendor if they buy a few Apollos
(or Suns). Or will they be able to use there old machines with new
machines out from XYZ with out respending a fortune?
Thanks
Todd Olson
ARPA: olson@lasspvax -- or -- olson%lasspvax.tn.cornell.edu@cu-arpa
UUCP: {ihnp4,allegra,...}!cornell!lasspvax!olson
US Mail: Dept Physics, Clark Hall, Cornell University,
Ithaca, New York 14853-2501