Ken Olsen, the founder of minicomputer and client/server company Digital Equipment Corporation (DEC) died on Sunday. He was 84 years-old. Olsen started out a maverick, pioneered and drove the minicomputer and supermini revolutions, and then became a dinosaur. But unlike many other senior DEC executives he remains a much-loved …
I'm only in my mid-thirtires, but I remember he first time I saw a pair of clustered DEC mini's swap a live Oracle DB instance between them without a blip and carry on running! Blew me away! Made up my mind to stick to the back-room stuff, it's way more fun than all that Windows GUI stuff that was taking off at the time, I knew Windows was never going to be as much fun as Unix systems!
... the photo brought back memories of operating one of these at what was the SURRC at East Kilbride when I was a post grad. All those switches to play with......
As a VMS junkie of some twenty-five years standing I have to salute you for keeping me gainfully employed since ceasing to be a Penniless Student Oaf all those years aago.
May he rest in peace
Obligatory PDP joke:
Q: How do you recognise yourself as a dinosaur.
A: When you know that big endian and little endian are not the only endians out there.
A sad day for those of us who cut our teeth on the PDP.
RIP Ken & DEC
We had PDP-11/34 computers when I started some 25 years ago, though they were replaced by PCs and Sun workstations by the 1990s. We had stuff hung off the CPU bus and still marvel as watching the huge HDD units working away, for all thier minimal storage by modern standards.
VMS was rock solid from what I know, and its a real shame they did not do as well as they could have. Also a real shame how "HP" dropped the Alpha processor (as Itanium was going to rule, eh?) and generally dumbed down all they had acquired.
Rock solid is right. On the very rare occasions that our VAX 11/750 fell over, Digital used to be sent the system dump on half inch tape* so they could analyse the problem themselves. You don't get that level of support from Microsoft!
Talking of Microsoft.. there is a great deal of VMS influence in the Windows NT architecture because MS hired developers from DEC to built an enterprise-class OS (for example Dave Cutler, http://en.wikipedia.org/wiki/Dave_Cutler).
So, although Ken Olsen is gone and his company is a distant memory, the work it did lives on in a variety of significant technologies across the IT field.
* which of course were over 10 inches across which caused confusion with the uninitiated.
Did the ex-DEC developers not also port NT to the Alpha platform, before MS decided to go Intel only?
Indeed. There's a whole hardware abstraction layer (HAL.. hahah) built into the NT architecture for this. Used to run on MIPS too, and MS are also porting Windows 7 to ARM, so again the HAL will be helpful in that.
"MS hired developers from DEC to built an enterprise-class OS "
Shamed they failed miserably. Win2000 onwards were ok , but NT 3.11 and 3.5* were lame dogs and should never have been released. Compared to VMS and the unix opposition at the time they were frankly a joke. The only "enterprise" involved was that of the MS marketing departments creativity in trying to persuade companies that these hopeless OS's were Big Iron replacements.
Re: @conrad longmore
"Shamed they failed miserably."
But they didn't. NT was replacing Windows 9x systems, and amongst programmers the sighs of relief were audible across the globe, despite the fact that it was regarded as a massive memory hog at the time.
No, it didn't compare to VMS and Unix, but it wasn't meant to - the hardware the PC manufacturers had at its release (the 80486 chips were only out for less than a year before) were not in the same ballpark as the minicomputer chips, although that gap was narrowing.
Re: Rock Solid
I.B.M ==> HAL (2001)
VMS ==> WNT .. a coincidence?
Sorry Dave but I can't let you run that <cut to WNT blue screen or red ring of death on xbox>.
Alpha was dead before HP bought compaq.
Compaq was already porting Tru64/VMS to "itanium". I was there.
"rainbow ... an appallingly bad product "
As obits go, that wasn't bad.
Rainbow wasn't an appallingly bad product, but it was ill timed; designed to support both Z80 and 8086 transparently, because at that time it wasn't clear whether the market for CP/M had been obliterated by the upstart MS-DOS. You can see why it might have seemed like a good idea at one stage, especially as to reduce costs the same enclosures and such were also used for a PDP11 (the Pro 320/350/380 family) and (if I remember rightly) for a PDP8 (DECmate III dedicated word processor).
Ken, you left lots of good stuff to be remembered by. Thank you. Raise a (non-alcoholic?) glass in memory of Mr Olsen. And/or a grenade for Mr Palmer and his successors.
Yeah, I agree: it wasn't a bad product, but the timing was wrong/unlucky. At college I used them in preference to an IBM PC because of the integration to the college 11/750 which meant I could use Wordstar to edit my Cobol code... Still have a lot of affection for the kit. My first job was at a Sperry shop and when you saw the error output from their compiler you sure appreciated DECs... The Pathworks and integration inspired me, and I soon got int o Networking, and that's why I do what I do today... Thanks Mr Olsen, I never met you, but you changed my life.
Ah, the Rainbow!
When working in the computing consultancy dept of a big multinational in the early '80s, we borrowed a Rainbox from DEC for evaluation. It was "OK" but nothing to get enthusiastic about... It was my first exposure to Microsoft application software - I discovered how to use a spreadsheet with MS Multiplan when my colleagues were promoting Visicalc to accounting depts around the world, then had to unlearn it when Lotus123 became the defacto standard.
By the time even DEC realised it was missing the boat on "personal computers", my colleague who was our liaison with DEC asked me to evaluate their latest offering: a Pro380. All I can remember of it was the superb (over-)engineering - it was built just like one of the minicomputers with heavy-gauge yellow-chromated steel chassis and panels and (ball-bearing?) slides for every adapter card, floppy and HDD. In the days when IBM were about to launch the PS/2 made mostly from structural plastics held together with press-studs, the cost of production must have been enormous - we never bought one!
I'm sure I won't be the only pedant pointing out that the PDP 8 wasn't an 8 bit machine, it was 12 bit (count those switches) address and data . This gave it a huge 4096 words (6Kbytes) of memory. (I think there was an expansion option to take it up to a whopping 8192 words)
The switches were necessary to load in the initial bootstrap code (RIM = Read In Mode?) which would then activate the teletype to read in the next stage - BIN from paper tape, which could then load the "high-level" interpretive language FOCAL. This was when computers really did pull themselves up by their own bootstraps.
DEC machines mapped out my start in computers.
A PDP 8/S (S for serial bus) was the first computer I had real hands on experience with programming, then I learnt a lot about computer hardware fixing a broken PDP/8E (blown transistor in the programme counter register IIRC).
Those bought up on modern hardware might be surprised to know that this wasn't a stack based machine, so recursion wasn't possible. The return address for a subroutine was stored in the first word of the subrutine, which always had to be left free. A return from subroutine was a jump indirect via this address. (this may not be 100% accurate - it was a long time ago)
The only storage we had was paper tape, and a fast - 1000cps - reader.
When PDP 11s came in, they were only for use by Postgrads, poor technicians like me weren't allowed to touch, though I did develop the A-D convertor to allow them to digitise "Scotty, Beam me up", which almost filled a RK05 diskpack (no MP3 in those days)
Later I went on to Vaxes, and loved them, but by then I was building my own microprocessor based home computers and the future direction was clear, but I owe my career path from analogue hardware technician to software developer, to Ken Olsen and DEC computers.
RIP Ken, RIP DEC.
And I forgot to mention...
VT52 and VT100 terminal protocols that existed long after the terminals themselves, even being used for elementary graphics on early PCs under the ANSI.sys driver and for Unix/Linux terminal emulators.
Not forgetting the DECnet Father Christmas worm, which was one of the first true network worms in the wild.
Good at their specialist area
My career grew in parallel with VMS 3.x through to OpenVMS 7.3.
A very stable & secure operating system, first with clustering that worked (AFAIK).
Loved the consistency of command line for system management jobs.
First with affordable 64 bit processors (Alpha).
The email system was ALL-IN-1 (no other alternative spelling!), and the underlying messaging system was and probably still is better than any MS offering (regarding message tracing especially).
I remember there being a Notes Conf (VAX..., not Lotus...) that listed all the alternate spellings of ALL-IN-1 and what they actually were. And the flame-fest that broke out when you used the wrong one. I think "All-In-One" in this article might have been a disposable diaper.... But then, I was in engineering so we never used it anyway :-)
Those were the days
I left school in the late 80's to write software for local government in VAX Basic, running on VMS5.5. It was such a good environment to cut my teeth on. Things just made sense, and were yet highly powerful. Happy memories of shelves full of bright orange manuals that would fold across the centre, thus turning themselves into book-holders.
One day whilst the Sysadmin was on holiday, the MD asked me (was only a small company), "Would you have a problem if he didn't come back?". And that was the start of my career as a Sysadmin.
Next job (mid-90's) was an aircraft maintenance company based at Stansted Airport. They had VAX's, a new application I'd never heard of called 'Pathworks', and a fair amount of desktop PC's running the new-fangled Windows 95 (using DecNET). Here, I learnt to deploy TCP/IP, to code HTML, and all about the Internet/email.
98-99 and I was at EDS (hated it) supporting massive OVMS-based systems. Initially using a DEC product called TeMIP - I've yet to meet another person (including HP staff) that have heard of it.
99-05 was my time at Made For Idiots - by this time, my OVMS usage was dwindling, and my Windows was increasing.
05-present and I'm still a Sysadmin, working for a major UK baby/mother retailer. There's no OVMS at all in my life. Instead, it's all Windows & DB's. There are some odd, green screen IBM things. But, I can't for the life of me understand them. I do have some Itanium-based Windows servers, which I gather have some/part of their basis in the wonderful 'Alpha' CPU.
My colleagues will testify that I now say (of Microsoft Clustering), "It was never this bad under VMS. It just worked."
Ken, I never met you. Never knew you. But, I feel you help put me where I am today. And for that, I am eternally grateful. Every day I worked with your products was a good day. I owe you a beer.
> A sad day for those of us who cut our teeth on the PDP.
while not a PDP, I started off in that era when we had a Data General minicomputer at school and I used to be able to program in the initial bootstrap via the front panel switches from memory - and it wasn't just the "green screens" that you used to access them ... there were teletypes as well!
I also rember, probably 20 years ago, listening to a radio program featuring DEC and Ken Olsen where one of the things I remember was Ken's (at least initial) insistence that his sales force were paid a standard salary and not commission on the basis that he wanted them to sell what was going to be best for the customer and not oversell to inflate the salespersons commission.
Death of an era
The PDP-11 is the machine that UNIX and C were originally written for.
And don't just gloss over that ...
There probably would have been Unix on some minicomputer, but the PDP was there first. Unix on the PDP/11 was followed by Unix on the VAX, then Sun machines (understand that I'm skipping things here a lot) then the PC and then with the licensing problems, Linux on a PC and now in my pocket. Well, not my pocket, but in the pocket of anyone who has an Android phone or any of the dozens of phones running Linux.
Ken Olsen called Unix "snake oil" while I was working at DEC (on Ultrix) but Unix on the PDP11 is what put Unix in the hands of countless undergraduates (me included) and the line from there, to well, an Android phone is easy to draw.
Writing for PDP-11
The PDP-11 was a joy to write assembler-level code (and MACRO-11 was only surpassed by MACRO-32 for the VAX). With an almost-orthogonal instruction set it would these days be a RISC machine -- unlike the VAX and the PC, the '11 had few instructions and a broad set of addressing modes and flexible registers to allow efficient implementation. Like:
MOV @(PC)+, @(PC)+
; rest of program, testing flag for zero/non-zero to see which entry point was used
At the time it was probably the best thing on which to write Unix (and the C compiler).
Indeed, its architecture is still reflected in C and C++, in the ways to increment a variable:
ADD #1,x (x+=1)
MOV x,R0; ADD #1,R0; MOV R0,x (x=x+1)
What I liked
was that the Program Counter was basically register 8, and that many of the jump, load immediate and return instructions were just special cases of other load and store instructions that you would use on other registers.
At one time, I used to be able to dis-assemble PDP-11 machine code without the book. It really was just such a regular instruction set that it was easy.
It used to be interesting to see just how C mapped into PDP-11 machine code. Often, like the case Keris quotes, a simple instruction like i++; would map into a single instruction.
Another innovation in reasonably priced computers that I believe was championed (although not invented) by DEC was the segmented address space that was implemented in such a way as to make it non-intrusive to the program writer, but also would allow different processes to have their own virtual address space independent from the physical memory addressing. It was this feature more than anything else that allowed multi-user computers to be created that allowed a user to screw up their own program without affecting the OS or other user's programs.
A ground breaking design
And I/O was memory mapped, completely different from the way we designed things in traditional mainframes. We bought 2 for a special project and concealed them in our trad blue boxes, much to the customer's annoyance. It was, I have to say, very difficult getting them passed by our QC people since they were so cost reduced compared to our normal stuff.
But they worked and were lovely to play with.
I join those mourning Mr Olsen as one of the real fathers of modern computing.
I may not have much experience with the meaty hardware, except epic tales I read about the DEC PDPs. My first foray into PC computing as a youngster was when my parents for me a DECpc LPV+ 433sx for homework.
It was a great machine, slimline factor wasn't as bulky as most PCs, was more business-like than a peecee world packard bell special, and surprisingly quick for a 33mhz 486 SX. Came with WFW 3.11, but managed to shoehorn Win95, a soundcard, a double speed CD-ROM and 12 whole megabytes of RAM in! Thems the days :)
RIP Ken Olsen.
The last of the great founding fathers?
RIP, it is a sad loss.
Once engineers set yo and ran computer companies and they engineered great products.
Now they've all gone and we are left with bean counters and marketters who view engineers with disdain even if they can't live without them, quite.
I severed my time on Ken's products. I programmed PDP11-23s, I even wrote some VAX assembler. I learned my Unix on Ultrix 1.0 on a VAX 11/780.
With Ken Olsen passing away a whole age of the computer industry passes too.
last of the great founding fathers - not quite
@Dazed and Confused - Gordon Bell (76) & Gordon Moore (82) are still alive and kicking.
PDP-8, 11 and Vax
Just for the record the PDP-8 used 12 bit words (and also had a sort of 24bit floating point capability with an optional expensive hardware accelerator). You could run real scientific programs in 4k, 8k or 12k words of memory.
Though DEC saw their biggest success with Vax and whilst it was an impressive architecture it came a bit late to the market. It was also encumbered by VMS which struck me as unnecessarily complex (being based on RSX/11) in contrast to the much more cleanly architected Vax Unix that DEC could have picked up from UC Berkeley a year or two later, but instead vacilated over for years.
One of Ken Olsen's less perceptive quotes ...
Someone has to make a reference to 'Snake Oil' - but I'll leave it to other to complete the quotation attributed to Ken Olsen.
Ken was lambasted in the trade press and by the competition for being misquoted. Unix was the up and coming thing; making its way out of academia into commercial use. Ken is stated to have preferred the undoubted reliability of the home-grown VMS and was reported as saying Unix is so much snake oil.
The full quote was that Unix of itself was so much snake-oil, by which he meant it was not a silver bullet to all problems of cost, performance, application availability. To that, I think he was right.
but his blind spot of not seeing the need for a computer in every home was a real blind spot.
I still have callouses on my fingers from keying in the bootstrap!
Two things stand out about both the PDP8 and 11, of which I have an 8 in storage somewhere, was that they were built like the proverbial brick sh*thouse and defective components could even be changed by a technician.
You could actually 'scope' an IC pin and watch a flip-flop change state - try that on those big, black blobs nowadays. A great technical teacjing computer.
Another big thing was interconnectivity! You could buy PC boards that would provide interfacing for almost anything - the PDP series were used by credit card companies in the early days for telephone line input at which they excelled.
Phillips, amongst others, produced a chipset for both the 8 and 11 that ran DEC software - none of the Apple crap in those days.
Talking about the California fruits, DEC had one hell of an App library that saved many a programmers neck.
Thanks, Ken, in spite of my callouses!
Good 'ol days
I too cut my teeth on a PDP 8. It was the size of three wardrobes, had two mag tapes, a hard disk and hi-res graphics (A total of 600 spots on a B&W CRT). Booting consisted of entering (40 odd ?) binary numbers on a toggle panel (loads of really cool flashing lights) to be able to read a paper tape, then load a paper tape to get instructions off the mag tape, finally load a mag tape, and after an inordinate amount of clicking and whirring you were able to start using the teletype. All long term storage was on paper or mag tape. The disk only held a few k and that was used for the focal interpreter.
It really did look quite impressive when both mag tapes, the teletype and all those flashing lights were going 19 to the dozen.
But you tell the young people today, ...
And another thing
Some years later I found myself at University and learned C & Unix on DEC Ultrix personal workstations, so I suppose I owe quite a lot to Ken Olsen and ilk.
Raise a glass to his memory !
KO was a brilliant, if flawed, genius
When I joined DEC in 1983, I was amazed to find a company culture like no other. Based on real family values, trust and such mottos as "do the right thing" and "it's easier to do and apologise than ask permission first". The engineers and innovators ran DEC at it's most successful and KO was indeed the "Ultimate Entrepeneur".
I was recruited to support the Rainbow because it ran CP/M and Wordstar and no-one at DEC knew anything about them. Of course it ran MS-DOS as well, but that was just the start of the confusion.
One of my customers was Douglas Adams, pre-Apple-obsession!
Even when the DEC "family" was over 100,000 strong, many people all over the world loved working for his company, which was innovation-led and often took the customer by the nose and prodded them into using a technique or process or machine that they'd never have thought they wanted and proving that they needed it all along.
Sadly, it all fell apart in the late 1980's. I remember being in the Littleton, Mass. facility and hearing that the stock price, which had touched on 180, was plunging as the fiscal reality of UNIX becoming the panacea of the industry and DECnet (a truly advanced networking stack in it's time) gave way to TCP/IP. KO had indeed led the company from birth into a decline from which there was no recovery.
He backed DEC technologies against the world, preferring VAX/VMS to UNIX abd even refusing the opportunity to buy the rights to UNIX when Bell Labs wanted to sell it. What a different world THAT would have been.
The innovations I saw at DEC still reverberate today: WAN, ethernet LAN, email, social networking (VAX Notes), clustering (VMS clusters were and are the best in class) Volume Shadowing (mirroring), database technologies such as two-phase commit, matrix management and so much more. Even the dreaded "Phase Review process" for development of hardware and software products foreshadowed Six Sigma in many ways.
I met KO, didn't know him well, liked him immensely.
RIP, another giant is gone.
Was the dog's 2.13s.
Many a fond memory of my time in DEC, thanks Ken.
The EuroForum notes conferences are not forgotten. Though I thought it was 2.12 not 2.13?
VAX Notes was the best social networking product written to date. Compared to the gibbering racket of Web2.0 social networking or the nightmare flame wars of usenet, VAX Notes had a signal to noise ratio that will never be bettered.
My memory went long before VAX Notes did.
Re: VAX Notes
Blimey, EuroForum is a blast from the past: I still remember how much of my daily schedule that thing accounted for! I never did find the correct timesheet code for it though. :D I wonder if my archives are still readable after all this time...?
I wish I'd experienced DEC under Olson's caretaking: everyone had very fond memories of the company under his stewardship, which was in direct contrast to Palmer's reign. Still, Olsen will be remembered far longer than his successors, and quite rightly; a larger than life figure in every way.
The Joy Of VAX
I only had glancing contact with DEC's products when I was tasked with making them talk to ICL's VME boxes a long time back in the way back when. Similarly to ICL's products they seemed to have been constructed to make the programer's and and the engineer's lives easier in a way that seemed to pass other manufacturers by when I came into contact with their systems, so I can understand the affection in which these boxen are held.
With him passes another piece of the computer industry when it was led by techies rather than business brains, a sad day.
It is worth remembering DEC's cooperation with Xerox and Intel in giving us Ethernet.
Pathworks was for networking PCs with VMS servers. Based on LAN Manager, it was rather late to market but to my knowledge, was the first that allowed DOS/Windows and Apple machines to share the same VMS file shares and printers.
RIP: So long and thanks for the memory
I grew up on PDPs,VAXen from the age of 16
I was cobbling some networking demos with the 'new' VAXmates when I met Ken. I was ~20, took me by surprise as he marched right up, shook my hand started talking away about new faster processors etc to come, can't remember much of it as I was just a bit surprised. I was told I was lucky but at the time never really thought I what was doing would make such a impression.
Nice bloke.. willing to talk to anyone. RIP sir.
the old days have just died
I worked for DEC through the 80's at the V7 building behind Huntley and Palmer's in Reading, and at DEC Park.
pdp 8, 11 and VAX, Rainbow, Pro 350/380 various PSUs and VTs all at component level.
Definitely an engineers company, loved it :)
Brings back memories
We also used several PDP11s at the university. My master's project involved running thermal simulations on a VAX, until I realized my then new PC (80386 at 25MHz with Cyrix math co-processor) outperformed it.
Essex Uni was one of the first (the first?) to get it hands on a DEC-10 in 1970. What an eye-opener. It replaced an ICL 1909 and introduced a roomful of teletypes to replace the card punches!
Online computing became de-facto for me from that moment. When I eventually became a business planner at ICL I was given another (batch) 1900 to do the modelling. I quietly slipped out and bought a TRS-80 on expenses, lashed a 132pp lineprinter to it and produced reports that looked like they had been done on a mainframe. Got mentioned in despatches by the speed I could turn round work.
I kept the secret from them. So thanks Ken for changing my computer life.
Although it was known as a PDP-10
they were labelled as DEC System 10 (and the followup, DEC system 20) running TOPS-10 and TOPS-20 as the OS. Real DEC sysprogs called them things like KI and KL systems, after their processor types.
IIRC, they were a bit quirky, having a 36 bit word length, but introduced the concept of a cluster with a fast interconnect. They had a thing called the CI bus, which was like an extended MASBUS that allowed you to connect systems together, as well as to Hierarchical Storage Controllers (HSC's) which provided shared disk between the systems.
This was adapted to become the BI bus for VAXen, which paved the way for VAXCluster.
- Apple: We'll unleash OS X Yosemite beta on the MASSES on 24 July
- Pics It's Google HQ - the British one: Reg man snaps covert shots INSIDE London offices
- White? Male? You work in tech? Let us guess ... Twitter? We KNEW it!
- The END of the FONDLESLAB KINGS? Apple and Samsung have reason to FEAR
- Researcher sat on critical IE bugs for THREE YEARS