tbray@watsol.waterloo.edu (Tim Bray) (11/30/89)
In article <1989Nov27.144016.23181@jarvis.csri.toronto.edu> jdd@db.toronto.edu (John DiMarco) writes:
A centralized authority -- if it is responsive to the needs of its users --
has the capability to offer better facilities and support at a lower price.
Just consider some of the relevant issues:
Resource duplication: ...
Maximum single-point usage: ...
Security: ...
Expertise: ...
Emergencies: ...
Backups: ...
Complexity: ...
Downtime: ...
Maintenance: ...
Compatibility: ...
Yeah, but politics wins, just like Barry says. Recently in Dig. Review, there
was a big fluff-article about client-server computing. No technical info to
speak of, but expert after expert brought up this issue of the intense
desire of users to be out from under the centralized org. To quote almost
verbatim: "Joe User will no longer tolerate interference from a bunch of
people whose job it is to keep things from getting done".
You might argue that one reason that MIS groups get such a bad rap is that
they have historically been purveyors of bad technology, i.e. IBM. But I
think the big reason is that they suffer from the same timesharing problems
that the big machines do. There are never enough of them to go around, so
the users end up waiting and pissed off. Simply put, there are very few
groups in the world that have the organizational and technical skills to
keep a variety of complex OSes and network facilities running, while
simultaneously dealing politely with the ignorant, all while seriously
overworked. It can be done, but it's hard.
lindsay@MATHOM.GANDALF.CS.CMU.EDU (Donald Lindsay) (11/30/89)
The Great Secret behind the success of minicomputers (and now microcomputers), is the discretionary spending limit. Many people are allowed to authorize purchases that are below a certain limit. Above that, the purchase must be first reported up through official channels for review. So, back in the minicomputer days, DEC found a lot of (say) $15,000 upgrades purchased as two $8,000 upgrades, simply because the customer's discretionary limit was $9,999. And some PDP's snuck in as "programmed data processors" (no kidding). Many early microsystems were semiworthless toys, incapable of solving the purchaser's problem. However, the total purchase was below just about everyone's discretionary spending limits. Even if management had forbidden computer purchases, the things snuck in as "office equipment" or "word processors". This is where today's industry came from. Believe it. -- Don D.C.Lindsay Carnegie Mellon Computer Science
linimon@attctc.Dallas.TX.US (Mark Linimon) (12/01/89)
In article <18794@watdragon.waterloo.edu> tbray@watsol.waterloo.edu (Tim Bray) writes: >I think the big reason [that MIS groups get such a bad rap] >is that they suffer from the same timesharing problems >that the big machines do. There are never enough of them to go around, so >the users end up waiting and pissed off. Simply put, there are very few >groups in the world that have the organizational and technical skills to >keep a variety of complex OSes and network facilities running, while >simultaneously dealing politely with the ignorant, all while seriously >overworked. It can be done, but it's hard. And it's rare than organizations understand that these things Need To Be Done and Take Time. So they buy a number of *nix workstations, that clearly don't need any time spent on maintenance or system administration :-), and give them to end-users who don't always understand some of the issues involved. Training? What's that? We bought the whole manual set... Well, I didn't mean for this to come out as a flame, but I think it's a relatively common occurrence. There is administration cost incurred on any style of computing; you can pay for it centrally or distributed, and up-front or in lost time when you find out no one ever understood how to do backups... It's just a matter of educating everyone, management and technical folks alike, and I don't have any generic answer. Note; this has strayed sufficiently far from architecture that I have directed followups to comp.misc. Mark Linimon linimon@attctc
mike@ists.ists.ca (Mike Clarkson) (12/07/89)
>From: lindsay@MATHOM.GANDALF.CS.CMU.EDU (Donald Lindsay) > The Great Secret behind the success of minicomputers (and now > microcomputers), is the discretionary spending limit. > > Many people are allowed to authorize purchases that are below a > certain limit. Above that, the purchase must be first reported up > through official channels for review. So, back in the minicomputer > days, DEC found a lot of (say) $15,000 upgrades purchased as two > $8,000 upgrades, simply because the customer's discretionary limit > was $9,999. And some PDP's snuck in as "programmed data processors" > (no kidding). This is so true. Where I worked previously, we bought one of the first Vax 11/780's in Canada by buying a "Level 3 Data Extrapolator" from a consenting third party. All to get around the reigning MIS dept. > Many early microsystems were semiworthless toys, incapable of solving > the purchaser's problem. However, the total purchase was below just > about everyone's discretionary spending limits. Even if management > had forbidden computer purchases, the things snuck in as "office > equipment" or "word processors". We even snuck our word-processors in that way too. There was a central typing pool that by decree we were supposed to use. The average turn around from the pool was 4-5 days. We were allowed to have secretaries, and you guessed it, "Level 1 Data Extrapolators", and all of a sudden we had 3 hour turn around. Which meant we could bury "them" in paper faster than they could bury us. Sort of like the advent of the Gattling gun. > This is where today's industry came from. Believe it. It's still there. -- Don D.C.Lindsay Carnegie Mellon Computer Science -- Mike Clarkson mike@ists.ists.ca Institute for Space and Terrestrial Science uunet!attcan!ists!mike York University, North York, Ontario, FORTRAN - just say no. CANADA M3J 1P3 +1 (416) 736-5611