Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!mips!pacbell.com!att!ucbvax!PAN.SSEC.HONEYWELL.COM!thompson From: thompson@PAN.SSEC.HONEYWELL.COM (John Thompson) Newsgroups: comp.sys.apollo Subject: re: HP Hype (new role for ADUS) Message-ID: <9105010226.AA02367@pan.ssec.honeywell.com> Date: 1 May 91 02:26:08 GMT Sender: daemon@ucbvax.BERKELEY.EDU Organization: The Internet Lines: 128 > The PRISM II upgrade is dead! No other company trying to build the chip - no > nothing!! If the answer of any of the HP reps there was honest, then they were > quite uninformed to say the least, since within HP this situation is known for a > while. They even appointed a guy (Ken Goldner - DN10k program manager) to ex- > clusively handle the task of dealing with affected customers. I doubt they were uninformed. Now that it's official bible, I can talk a little more freely. Almost two months ago, my sales rep called me to take a survey of what DN10000 equipment we had (all on maint, but let's make sure), and what we were running on there -- Apollo tools, 3rd-party tools, In-house tools. He said that the 2X upgrade was dead, since the FPU had problems, and there would be at least a 6 month delay (killing the window of opportunity). When I was about ready to mention this little tidbit, I decided to call up my rep again just to confirm it. At that point, he said that he had spoken rashly, and that "no decision has been made -- we're investigating all the options." Because of this, I'd tend to accuse HP/Apollo of deliberately giving a snow job. > Now what? I have a nice DN10020 with lots of RAM and diskspace (of which I > can't use any in other DN-type boxes, right?), The RAM is definitely custom. The disks can be put into the DN3550 and up, assuming you have the WDC-7000 controller. (ACtually, if you have the Maxtor 348MB disks (XT-4380E), you can put them into DN3000s and up, with the right controller (Omti?). Of course, those nodes are _really_ dead! > which is basically as fast > as a Sparcstation 2, or should I say as slow, given the performance figures of > the Series 700 in comparison. I wouldn't be quite so hard on it. The DN10000 is still a pretty fast box, and with the multiple processors, I feel it's still merely comatose. We are, however, looking at eventually using our nice shiny DN10000s as file-servers! > We had ordered our PRISM II upgrades in November 1990. HP now offers, to take > the amount comitted for the DN10000X cpu boards, and carry it over into a pur- > chase of a 750 (or even two) at significant discounts (40% and then your end- > user discount == 40% + 24% --> 54.4%). Example: We had ordered 3 10kX boards for > $47k - we can now use these $ 47k to buy Series 750 stuff at the specified dis- > count. Not a bad discount. > That's nice, but what do I do with a 750, if it doesn't fit in my Token-Ring > network running Domain/OS yet, if it will not run OSF/1 until November or > December of 1991, which means if I get it I have to support yet another pro- > prietary operating system in HP-UX??? BINGO!!! > It is like: Me, trying to buy two more apples to add to my fruitsalad, and HP > says: sorry, we don't have them, but we can give you 5 potatoes instead! It's worse. They said: "Sure, we have apples. Let me go to the back room and get them for you. <5 months later> Apples? Oh, no, you really want potatoes." > Well, that's one way of trying to ease the pain of Apollo users. The other, > and YES, I still haven't given up on it completely is, to try to convince HP, > that the Series 700 will only be a full success, if Domain/OS will run on it! > Us Apollo users deserve not to be left behind. And OSF/1 is no answer yet. > OSF/1 is too far into the future - realistically most people will not jump on > it before version 2.0. They claim Domain/OS is not dead yet, but they are not > running it on their best available computer? There is, unfortunately, a problem with this. Even if they put a _LOT_ of effort into it, there's no way that there'd be a stable HP-PA/DomainOS up and tested in under a year (IMHO), unless they've already started. Given that they haven't started (I tend to expect lies from them that'll hurt, not help) then it wouldn't be available until they're at a Domain/OS that'll hook into the OSF -- Q3-92. There is no reason for them to waste (yes -- waste) their time on a dead horse. The big problem is that upper HP management decided early on to screw the Apollo people, and they held to it. > So I guess it is dead, then, isn't it? Yup. Now, just learn HPs-UX (or is that HP-sUX?) > >Thus, ADUS has an educational mission to its members, to explain the simple > >economics that mandate a rapid change to OSF (yeah, I know it isn't even > >shipping yet, but "rapid" to me means 2 or 3 years instead of 4-6). Rapid has never meant 2-3 years in the workstation field. I will go to OSF as soon as it's out the door and we have 1 (any 1) product to put on it. That includes Mentor (and everyone they've bought out) Cadence (and everyone they've bought out) Synopsys (and have they bought out anyone) in-house tools (if we're not bought out :-) Unfortunately, I may even dip my hands into HP-UX. The market drives us, and the snake boxes are certainly fast. We expect to see several of our vendors with an HP-UX port this year. > Well, my definition of "rapid" changed last Friday, when the HPollo sales guy > told me, that our $100000 (just hw) investment in our DN10020 became obsolete > in less than two years! That's rapid, indeed! Simple economics, eh? A (tiny bit) harsh. The DN10000 has been available now for about 2-1/2 years; you shouldn't really count from when you bought it. (If we did that, we'd have a system that is obsolete 4 months after we bought it!) Of course, we bought it with a fair amount of knowledge about the deadness of it, but it still remains the fastest Domain/OS box around, and we need a lot of Domain features for our tools. > >Anybody have any other ideas? :-) > No ideas - just shaking my head. It's (unfortunately) not going to be possible to get Domain/OS on the snakes. Let's try for this instead. Get OSF up and __STABLE__ right away, if not sooner. This includes token-ring support. I am _not_ going to be able to get my building re-wired! Get the Domain/OS hooks for OSF in right now. If it needs to be done in stages, then so be it. We need to have shared registries. If OSF can handle NFS as well as AFS, we _could_ get by with NFS X-mounts for a while, but we'll need real file-sharing soon. The NCS hooks should come in with the registries, but if not, we'll need that next. If I can have that, I'll bitch and moan, but I'll be able to live with the box in my network. Get OSF on the snakes into the 3rd-party people NOW!!! I'm sick of hearing how easy it is to port to HP-PA and HP-UX. You can have all 2100 applications on that little box, but I won't be happy with the system until I can incorporate it into my network. An HP-UX box is as much a part of my network as our Suns and Vaxes (i.e. not at all). Do _NOT_ hold back on any of the above in order to make HP-UX a little nicer, either for us or for the HP-UX people. From a purely economic standpoint, you (HP/Apollo) have a _LOT_ to gain from getting us to move to the snakes. A lot of us won't move without a transition path. Jumping to a proprietary, obsolete, brain-dead O/S is not a transition. -- jt -- John Thompson (I'M ENGAGED!!!!) Honeywell, SSEC Plymouth, MN 55441 thompson@pan.ssec.honeywell.com A pessimist sees the tunnel. An optimist sees the light at the end of the tunnel. The realist sees the tunnel, the light at the end of the tunnel, and realizes that it's an oncoming train.