The midseventies SHARE survey 64

Lots of snippage below

is

Founded in 1955 as a self-help organization of scientific and education customers (recall that in those days there was nothing *but* mainframes). The similar organization GUIDE served that purpose for IBM mainframe shops more oriented to business applications.

SHARE had been around for almost a decade when the S-360 was first announced.

I don't recall if my response was in that file, but if you find anything attributed to SHARE installation code UNT, that's me.

Why is that a surprise? Today it's a given that computer technology advances at breakneck speed and that by the time a new computer reaches the market it's already obsolete, but until very recently the enormous price tag of computers -- even a medium-sized installation would easily cost in the millions -- and the disruption of replacing a machine meant that management, with good justification, would expect a decade or more time in service for a computer.

As a data point, in 1968 plus 1 I bought 256 KB main memory for a 360-65. It cost me about $330,000 after discounts. Plus shipping. Plus maintenance.

Besides, the S-360 line was announced in 1964, and it wasn't until 1967 or thereabouts (I'm not sure where my old announcement letter fiche is hiding) that a viable release of MVT showed up, and even after the S-370s were announced MVS didn't arrive until even later.

...with the non-virtual-capable 370-155 and165. In his postings Lynn and others have described these machines in much more polite language than I would have used.

There were also several 360-91 shops; one consequence of this was that the SHARE community became aware of the LSPS enhancements for OS-360 that IBM originally made available only to the91 shops, even though those changes offered huge performance improvements for all OS-360 users. If anyone here recalls the "Let my LSPS go!" signs on the Paddle, I recall sitting in Robert Rannie's hotel room lettering those signs.

As noted above, GUIDE was the organization of business users, although some shops belonged to both organizations. SHARE membership requirements included owning, leasing, or running an IBM-nameplate binary-capable machine; I don't think that any character-oriented box qualified. IIRC, by the mid- to late- '60s this meant a 704x, 709x, or S-360-50 or better.

Satisfaction with third-party hardware was in many cases utterly dependent on the quality of the maintenance staff, and the willingness of the vendor's field reps to go to bat for the customer. That's not to say that every IBM rep was perfect; one other shop that was handled by the branch office that owned our account had a problem with one of the CEs who was a whiz on accounting machines -- if you had a broken 402 or 403 you *really* wanted this guy to show up -- but he was equipment back."

But...in general, the IBM support was *excellent* and in almost every shop I encountered had a very good working relationship with the customer staff. (When I left my PPOE the CE staff from the local IBM branch office took me out to lunch.)

Third-party support was a real problem, and some of the design blunders in the design of the OEM boxes didn't help.

In this time frame, lots of the OS-VS2 shops would have been running SVS, not MVS. Admittedly, SVS ("Single Virtual Storage") was essentially MVT with virtual memory grafted on and the product was intended as a bridge for users moving into MVS, but MVS was still a big step. (Perhaps Lynn or others can provide a timeline for SVS vs. MVS.)

H'mmm...lessee..."improve the performance and reliability of the OS". Somehow that sounds like what we're wanting Microsoft to deliver.

Some things never change. (Disclosure: I was for a while the manager of the SVS project in SHARE, and later folded it into the so-called Paddle Project which had as one of its deliverables the Debt Collection of user- written patches and enhancements.)

I would be very surprised if any multi-vendor shop avoided this.

You're talking about a world thirty years in the past, when (for example) memory implemented in silicon was just beginning to become economically feasible. The 256 KB memory I cited earlier in this response was a cube about 9" on a side, and just to service that cube required a cabinet about 7'x 6' x 30", filled with power supplies and electronics. Microprocessors were in the future (at least for the general population), meaning that there was no way to avoid the delays inherent in having the CPU spread out over a large physical space with nontrivial signal propagation times between components.

Joe Morris