-----Original Message----- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Alan Altmark Sent: Friday, September 08, 2006 9:54 AM To: IBM-MAIN@BAMA.UA.EDU Subject: Re: FW: Fatuities (was 'Another BIG mainframe . . . ') <snip>
Jon, I think z/OS sysprogs should continue operating their z/OS systems just as their users have come to expect and their business requires. That's where the goodness starts. I feel that when a mainframe goes out the door it's because there wasn't sufficient "grass roots" support for it and that is because the *users* abandoned it in favor of alternatives that are perceived as "good enough". The sysprogs don't make the "buy" decisions, but they *do* influence the opinions of users who *do* influence the buy decision. A bad sysprog will kill the mf straight away. A good sysprog gets the opportunity to *prove* that s/he can provide the same service for less, or better service for the same money. Empire-building is not dead. "I own 60 servers" vs. "I own a mainframe". Who has the power? <snip> Mainframes go out the door because snake oil sales people sell a company [Read that: Upper management] on GUI and wiz-bang this and that. So they are told how expensive the mainframe is, and they are shown charts of this and that. I had a client that was told that the mainframe could not do GUI, was green screen only. I was FORBIDDEN to turn on HTML in CICS to prove them wrong! I had another that flatly stated that nothing negative was to be said about the new system they were migrating to. No one was to point out the missed deadlines, or problems. Later it was declared a success. The CIO has "retired." What MANAGEMENT [the decision makers] are NOT shown by the sales people selling the new system is the hidden cost of ownership of the new hardware and software. They are NOT shown the TCO (total cost of Operations). The PRIMARY reason these things are not shown is because they are HIDDEN and are NOT KNOWN. What these companies are doing is getting a new system that is all GUI on cheaper hardware, and the design is really set up to emulate their current environment. If you want to emulate a mainframe environment, one of the things you have to know is, how many attach points are there in your applications? How much of your system is interconnected? Then get out your checkbook to pay for the communications gear you are going to need. Now, given that almost everything is somehow connected via cross-memory connections (please do not confuse this with actual DAS, SRB, etc.) and happens at MEMORY speeds, try to pull this off with GiGE or 100BaseT or whatever. Most of the systems that I have seen try to take over just ONE application from a mainframe (running under CICS) have fallen down because they have to have two or three machines running data base (with some kind of load balancer in front of them!), three or four doing web page service (again, with a load balancer), two-three applications servers. And the I/O load that they experience in interprocess communications [read that, box to box communications] is what drives them to their knees. The amount of network slowdown is incredible. There was a paper, wish I had a copy of it here at my office, where someone went to great lengths to demonstrate all of this mathematically. If their paper could be put into simple language and presented with management attention span sound bites taken into consideration, you would see more people sit up and take notice (Cynical? Not hardly! I talk with sales people who tell me that decision makers want simple statements, quick meetings. Too many do not want tech details, that's for their techs to handle.). So I'm not too surprised to see some company run off and throw out their mainframe. And then quietly have to bring it back in (some one else here rightly mentioned "face"). One such company was very involved in bringing you the North East US (and Canada) blackout. Management decision making that is geared toward bottom line, with no understanding as to the ramifications 18 months later. If only managers would look at Bill Carico's Reboot Hill site (http://www.actscorp.com/reboothill.htm). I've given them two examples to put there, don't know if he has. <\red vision mode> BTW - I keep telling managers, my job is to make IT BORING. Exciting is when things run faster for less than projected. Annoying is when you get a phone call at 2AM and the major application's check run has failed. Later, Steve Thompson ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html