18 Colburn Ln Hollis, NH 03049 July, 1997 Managers Digital Equipment Corp. Gentlemen: In spite of my wishes and rather strong prejudices to the contrary, I have come to the conclusion that it is no longer prudent to remain with Digital. The saddest thing, to my mind, about this is that the conditions which led me to this conclusion are conditions over which you folks have control, and which could be changed. I have been active in the industry for a good many years now and believe my crystal ball in these matters may however be more accurate than most. I will tell you up front that one of the more important conclusions is that you should stop killing off OpenVMS. This marginalizes everyone's efforts working with it, alienates customers, and will I believe kill off Digital before any alternative strategies have time to materialize, presuming they could work in any case. However, there is a larger picture you need to be given before I elaborate. In the hope that my advice may prove to give some benefit to Digital, I will offer it here. I have done so occasionally to some of you in the past, but perhaps notice in this form will prove more useful. As I examine the operating system market, I believe that the chances of breaking Microsoft away from Intel architecture are small but DEC isn't taking advantage of what could be done. Given these circumstances it appears to me that any dreams of making Alpha take over the current x86 role in NT are vain. Note that the IA64 architecture will be a new instruction set, so that all the current x86 software will have to be redone to run in the new native mode. That (coupled with the notice that DEC has experience with dual mode machines which shows rather clearly that the machines run at the slower instruction set speeds until essentially all code is the new native mode) could just possibly push the market to consider Alpha. This is not being done as far as I can see. However, the upcoming dislocation in the x86 world offers some chance that the picture can be changed in a larger way...if you are willing to stop being analysts and to start realizing you are in a position to change the shape of the future. I advise you to continue getting systems to work with NT, but to push two OSs that Microsoft doesn't control: VMS and Linux. VMS and Linux can benefit from better NT connectivity, but Linux is poised to take off in volume. Developers who at first believed that Microsoft was telling the truth about being open to expose internal interfaces for NT now know, in large numbers, that this was a pack of lies. They'll continue doing NT things as long as there is no large market alternative, but Linux is developing into one. Meanwhile, VMS has the potential (which the Galaxies effort will begin to exploit) to take over the high end. Let folks know what VMS can do and give some evidence of belief in it, and stop telling folks that NT will have everything VMS has in a year or three (ten maybe. Three, no way. And ten only if VMS stops.). The repeated statements about how NT is the future are incredibly harmful to VMS (and I believe make Digital's future quite shaky). Strive to build for a computing paradigm in which OVMS is the high end system that everyone can use for safe data storage, retrieval, and management. Then remind developers (who after all control where applications get developed) that VMS internals are fully open...listings are readily available and inexpensive for internals sources. Basically, grab the high end and push down. At the bottom end, Linux bids fair to eat Unix' lunch, and might wind up eating NT's dinner. Push that, and push use of Alpha in the Linux community with more than the unofficial efforts done thus far. Build connectivity and make it easy for Alpha Linux to talk to VMS and NT and to unix. Support every effort at running NT or Windows apps on Linux as well. You don't get much for the OS, but you may be able to push an incredible amount of iron, and VLM databases, mark my words, are just the beginning of useful tricks possible in a 64 bit address space which can't be done in 32 above toy scale. Push Alpha Linux at low end as a way for serious developers to do next generation apps where they are not address space limited. Intel won't follow for a while and NT is going to have grief moving up too...an installed base problem and what looks like a serious code entropy problem at Microsoft. VMS iron should stop costing more than Unix or NT iron; I've seen these OS killing efforts before...I recall, for instance, IAS... where prices are raised to kill off an OS. Most of the world sees the pricing as DEC's attempt to kill off VMS. Are they right? Even though the VMS development is directed at the high end, the abandonment of workstations is devastating for VMS. What, after all, are developers supposed to use? Or smaller shops? This is taken as yet another signal VMS is being killed off. Then too, it is really essential that DEC start using the word "VMS" again instead of doing foolish and asinine stunts (as at the recent DECUS) of asking if people's OS choices are NT, Unix, or Proprietary. (As if NT or DEC Unix are not proprietary!) Beyond that, the fact that VMS funding has been cut year after year, and continues to be cut, is a more clear indicator of Digital's corporate commitment to it than any statements anyone can make. Conversions claimed to the belief that VMS is strategic, when unaccompanied by any dollars, are not believable. If Digital can't or won't even admit that VMS exists, it's mighty hard to see how VMS (or for that matter Digital) can prosper. You're marginalizing VMS and everything that's done for it, where it has the best, bar none, clustering in the business, has the best, bar none, security in the business, and with the Galaxies effort will own the high ground in computing performance. That's a hell of a good story. Will anyone breathe a word about it? From what I can see, VMS and all who work on it are being made irrelevant, and their skills are being made irrelevant, due to policies that can and should change. I'd love to be able to keep doing wonderful things in VMS...but must consider my responsibilities to my family and my wishes to have my work used. I know I am not alone in such feelings. I have donated code to Digital which will solve the problem of having a filesystem approach suited to 10000 disks or more and have published some VMS security add-ins which give VMS some unique capabilities. They're great stuff, and I just hope that someone will benefit from them. I do however begin to feel that I'd need to concoct a program able to create life or spin straw to gold under VMS to get it noticed in the world at large, so well concealed is the VMS image. (I was flabbergasted a few months ago when I was asked what VMS was...by the editor of the Nashua Telegraph!!! Right in the town where VMS is built! Helluva secret going here, folks.) I ask then that you rethink your course here. Take off your "analyst" hats that are predicting that NT will take over the world, and be aware that by changing what the world knows, you can change the future, to the benefit of Digital and the delight of those of us who wish Digital well. And for God's sake, take some of the advice you have here, and some other which you've been getting for ages now from others in the company. While I will be doing other (and I believe quite significant) work shortly, I plan to remain active in DECUS as I have been these 25 years past or so and hope that the foregoing will be of some benefit to Digital, my colleagues, and the computing industry. You can make it so, gentlemen. Yours Glenn C. Everhart, PhD. DECUS OVMS SIG Librarian & Tapecopy Coordinator