It's good to hear that the VAX was a cost-effective solution - there are too many stories about how expensive DEC gear was, but I imagine they primarily came after PCs started dropping in price. On 9 February 2016 at 04:50, Ethan Dicks <ethan.dicks at gmail.com> wrote:

Any idea about the date of when VMS could do that with a VAX? I don't remember how expensive a Cyber 3300 was back in 1967, but I worked at Northern Electric in Ottawa at the time. The standard operating system was being used (I certainly can't remember the actual name) and there were CDC (washing machine sized) disk drives with those removable platters. The IT department produced software which was able to support many users editing files via CRT terminals. I can't remember the maximum number, but I would estimate that it was at least a dozen. There were also a number of commands, one of which allowed a file to be submitted as a batch job. The batch job queue was able to run ONE job at a time. There might have been a high priority queue for very short jobs. Every night, the system backed up all the hard disk files to tape. At one point, the size of the file directory was increased from about 2000 files to about 6000 files. In addition, the search algorithm changed from linear (start to when the file was found) to a hash value based on the name of the file - normally the first file access found the file label block number. The reason for this reply is to document that there were already such systems available with very innovative software solutions as far back as the 1970s. About a year later in 1968, a utility opened in Toronto, MAGCC (Multiple Access General Computer Corporation if I remember correctly) which supported the identical software using a Cyber 3500. Thus the software concept had started to spread, They used dial-up modems, probably around 300 baud. I don't remember dial-up modems being capable of 9600 baud until a decade later. I also can't remember what a Cyber 3300 cost back in 1967, but I would suggest that the cost per user was probably similar to what VMS on a VAX supported, but MANY years later than 1967 to 1968. If anyone is really interested. it might be possible to search for the information. However, Northern Electric is VERY long gone now! Jerome Fine

Indeed. RSTS/E did better, with less hardware -- 64 users on an 11/70 was no problem, and earlier on you could run 16 users on an 11/20 (though not all that comfortably). paul

In the mid-to-late 70?s, I was an undergrad at CMU. I had accounts on the CS department?s systems. When I started, it was 2 KA10s. Eventually a KL10 was added (all running a *heavily* patched version of TOPS10). During the day it was nearly impossible to get stuff done with ~200 faculty and grad students on-line (on each system). I also spent a fair amount of time on C.MMP but I only wrote code for it and never actually did any ?production? work on it (but it was *cool* to see?cross point switch was the ultimate in blinking lights!). TTFN - Guy

In late '86 I was running a VAX8500 with 2 RA81s and a TU80 for Deluxe Coachlines in Australia. We computerised their coach reservation system. System growth was brisk as the company and the application grew. The 8500 became an 8550 and gained some more RA81s and then in late 87 we move to a new site. Prior to moving a new 8550, 2 HSC50s, star coupler and 24 (count 'em!) RA81s in RAID 1 pairs, a spare and a quorum drive and two TU78s (master/slave I think) were installed at the new site. For the actual migration we just moved some data on tape. About a week before the move the live 8550 in the old site failed with a double CPU fault. I rang the field circus number, logged the fault and impressed upon them the urgency of the call. After too long with no sign of an engineer I called again and was told the engineers were on site and working on the fault. As I could look directly into the computer room from my desk I knew this was not so. I restarted the live 8550 and called field circus to vent. They'd got confused between our two sites and two 8550s and gone to the new site. Turns out that the new 8550 had a double CPU fault too! We decided to ensure the new 8550 was working and moved with little fuss a week later. We then moved the old 8550 and completed our VAXcluster. We later upgraded the HSC50s (took ages to boot off tape) to HDC70s which I think booted off floppy. Each 8550 ended up with 80MB RAM. It was a pretty big system. I left Australia in March '88 and visited again a couple of years later by which time an 8820 had joined the cluster. Thick Ethernet served the office using LAT terminal servers and two (maybe four) big terminal servers (PDP11 based think) were connected to Penril multiplexers for an Australia wide terminal network. Then there were the laser printers whose power on self test display included a kangeroo bouncing along the one line LCD display! Fun times. On 11 February 2016 at 21:41, Glen Slick <glen.slick at gmail.com> wrote: -- 4.4 > 5.4

My comments regarding RA81s run something like ?we purchased two of the first RA81s in Australia but ran about 15 of the first 100 shipped?. My FS guy and I could replace an RA81 HDA in 15 minutes - we had lots of practice - I think we replaced 10 or 15 in six months :-( Huw Davies | e-mail: Huw.Davies at kerberos.davies.net.au Melbourne | "If soccer was meant to be played in the Australia | air, the sky would be painted green"

There's a reason why we ran the RA81s in pairs in the cluster... On 12 February 2016 at 01:39, Jay West <jwest at classiccmp.org> wrote: -- 4.4 > 5.4

The most amazing timesharing system I've seen is PLATO. In 1977: 600 active users, highly interactive, on a pair of Cyber 73 (CDC 6500) dual CPU machines. Each CPU good for about 1-2 MIPS, so that's pretty good. Even with a customized workload, that was a tough lift; typically about half the keystrokes would require "real work". Not like a command line based system where most keystrokes just get echoed and stuffed into the current line buffer. paul

Since your answer seems to be an unstated question, I will attempt to answer what you might have inferred. I was just attempting to note that the CDC 3300 hardware was able to support that function more than a DECADE earlier. By 1967 when I arrived, the system had already been running for quite a while in support of many engineers who required access to a computer that avoided the punch card route followed by submitting the cards to the batch queue - and waiting - and waiting. Access to an editor to compose the file followed by immediate placement into the job queue was a huge improvement in the mid 1960s. Agreed! But when you discuss the VAX and VMS, that is a decade later. As long as the Cyber 3300 was able to get the job done in 1966, then a quibble over implementation more than 10 years earlier seems a bit silly. I did not have a question! You made your answer into a question with your TWO question marks. If by factory floor, you mean where the hardware was built, then based on 55 years of developing programs, I disagree. "Timesharing", in my opinion, is a result of the operating system software. For example, RT-11 running on a PDP-11/73 (or faster CPU) is certainly NOT designed to be a timesharing operating system, although under very specific conditions it can operate in that manner. TSX-Plus will execute on the identical hardware and have excellent performance as a true timesharing operating system. But the PDP-11 did not become the powerful computer that it eventually became until at least about 1980. When RSTS/E was combined with that level of hardware, it was able to support timesharing as well. At this point when it is possible to run RSTS/E on hardware that runs 100 times as fast as a PDP-11/93, staying with RSTS/E makes sense for applications that are still basically the same after 30 years. So, back to my original statement which comments on much praising of a VAX supporting editing sessions (along with running the programs which were produced) around the mid 1970s. I doubt that the original 1.0 VMS handled timesharing very well. So I don't believe it was unreasonable to make a statement that the same functions were available more than a DECADE previously on a Cyber 3300 even though, as you point out, the operating system in the mid 1960s was probably still at the level of RT-11 as opposed to TSX-Plus. I don't remember if the Cyber 3300 hardware even survived more than 10 years, but I suspect if it did, then by the mid 1970s it would have also evolved into something that was much better than 10 years earlier. Jerome Fine

Thanks for getting the case correct for DECsystem-10 and that other one :-) When I started managing VAX/VMS on an 11/780 we had two RM03 disk drives and 1.25MB of memory supporting 16 users. It was really important to get more than 1MB of memory to ensure that you got the larger memory boards that would allow expansion to 2 (4?) MB. ISTR that with educational discount our 11/780 cost about AUD$250K in 1982. I guess I?m lucky in that I still manage OpenVMS systems today 30+ years later? Huw Davies | e-mail: Huw.Davies at kerberos.davies.net.au Melbourne | "If soccer was meant to be played in the Australia | air, the sky would be painted green"

On Thu, Feb 11, 2016 at 9:56 AM, Mark Wickens <mark at wickensonline.co.uk> wrote: VAXen weren't exactly cheap (our 11/750 was ordered the week they were announced and was, I'm told, the first one shipped to the midwest (S/N BT000354). ISTR it was $120,000 with 2MB and 2X RK07, then we added a Systems Industries SI9900 SMD controller and had a 160MB Fuji drive and a 400+ MB Fuji Eagle. I don't have the numbers on that, but it was another $30,000 at least, I'm reasonably certain. I do know when we bought an RA81 in mid-1984, it was 424MB for $24,000. DEC VT100s were around $1,800 in the early-1980s, I'm pretty sure, then we switched to CiTOH terminals for around $1,600. When you add in 16-port Emulex CS21s to attach all those terminals, and divide the cost of the central machine by the number of users, I think you get that $5,000 per user cost. IBM-brand PCs were also about the same cost per seat, but the software was a lot cheaper (minicomputer apps were tens of thousands of dollars for a handful of users; PC apps were hundreds per user), and PC maintenance was way cheaper (but not as convenient as a full-service maintenance contract). So, yeah, DEC gear was expensive and PCs were cheaper, but perhaps not as much cheaper as people felt they were. BTW, these are all new prices from when the gear was first launched (when we bought those items). Where the savings came was if you could live 2-3 years behind the leading edge. I bought a lot of DEC equipment from resellers that kept us going for more than 10 years. Lots of other companies went with Ethernet and PCs on every desk, etc. We stuck with minicomputers (VAXen and PDP-11s) for the vast majority of our work (correspondence, software development, cutting customer tapes...) and used PCs for a couple of specific tasks (accounting, because Peachtree on an IBM PS/2 Model 30 was way cheaper than any equivalent app for VMS, and circuit board design with OrCAD and PADS-PCB). So there's a slice of the mid-80s and how we got things done. I'm reminded a bit of all this recently as I'm refurbing a DEC VT220 and an IBM 3130 terminal, to use with a modern Linux box to share with folks the experience of life on dumb terminals. I'll probably make a post soon, once I get the hardware all cleaned up, calling for favorite DEC and IBM games that play on dumb terminals. -ethan

Just thinking about Mark?s email here. Keeping in mind that this is from a DEC persons point of view. You probably all know and appreciate this. While the DEC hardware may have been more expensive than other companies many customers are still running the hardware today. I am thinking of several customers that are still running PDP11?s in a production environment. I am also thinking of VMS customers that have been up an running for 33 solid years. This is almost inconceivable with current hardware and software. When I worked with Nemonix we had customers needed replacement parts for the majority of the server hardware that DEC made. This was on the server side not necessarily the Network products like the Gigaswitch or hubs. Just my 2 cents, Have a great day, Sue Sue Skonetski VP of Customer Advocacy Sue.Skonetski at vmssoftware.com Office: +1 (978) 451-0116 Mobile: +1 (603) 494-9886 Connect VMS Boot Camp will be the week of Sept 25, 2016 at the Radisson in Nashua NH.