Slashdot Mirror
Windows OS Coming To the Mainframe
msmoriarty writes "Following up on its May announcement, IBM has now confirmed that by December 16 it will support Microsoft Windows on zEnterprise via its zBX component."
← Back to Stories (view on slashdot.org)
What in the Mainframe market sector is this the answer to?
Now when work calls me at 3AM because a mainframe job failed I will have to say "Please reboot the mainframe and call me back again if it still fails!" Then I can go back to sleep.
The summary misses something fairly important, which is that Windows isn't running on the z mainframe itself. This allows Windows blades to be inserted into an external chassis (zBX) and managed by a software component called the Unified Resource Manager.
From the article: "Make that Windows right next to the mainframe -- i.e., running on the zEnterprise BladeCenter Extension (zBX), the mainframe/open systems sidecar...First, Windows-in-a-zBX isn't Windows-in-zVM. Still less is it Windows running in a special processor, a la IBM's Integrated Facility for Linux (IFL). So Windows won't be running on non-x64 -- i.e., Big Iron -- CMOS. Nevertheless, customers will be able to manage Windows from their zEnterprise 196 or zEnterprise 114 mainframes...
Advice: on VPS providers
Reading through the thick IBM-specific jargon, zBX is actually a blade server management system that places blade servers on a private network connected to the mainframe, with the mainframe managing them. It supports POWER7 (FYI POWER is a "big cousin" to the PowerPC chip) and IBM System x (x86-based) blades.
So, in actuality, this is Windows running on an x86 box, with the mainframe managing it -- it is not like mainframe Linux where Linux is truly running on the mainframe.
... Ferrari have just announced that they will be installing a 125cc engine into their 458 Italia. A spokeman said: "It works'a fine in'a the moped, whats'a the problem?"
The Devil sits on his throne in Hell. On of his minions comes running in.
"Sire! Sire! Microsoft has ported Windows to a mainframe!"
The Devil favors him with a surprised look. "Is it that time already? The end of the mortal world?"
The minion genuflects before him. "Yes! Yes! End of times, master!"
The Devil rubs his chin. "Windows on a mainframe?"
The minion nods emphatically.
The Devil considers it for a few moments, "Well, I don't think I want it anymore."
Truly, next year will be the year of Windows on the Mainframe!
Mainframe market share is huge (about 3 billion a year in hardware sales) and growing rather quickly, especially since "cloud" and "virtualization" became buzzwords. It's the only part of the high-end non-x86 niche that's really having solid growth right now - SPARC and Itanium have been tumbling for a while, and Power has been more or less flat.
Only one of the last 6 companies I worked for DIDN'T have a mainframe.
Not only does my current company still have a mainframe- we're doing a major software upgrade on it next year.
The mainframe never died.
"That's the way to do it" - Punch
Some estimates have mainframes processing 80% of the world's data. http://fosspatents.blogspot.com/2010/08/western-civilization-runs-on-mainframe.html Now I'm not sure how accurate that percentage is, but if you run an enterprise business and have thousands of servers to maintain, a mainframe still makes a lot of sense.
Mainframes are increasingly seeing competition from clusters of commodity machines running Xen or similar - the cluster is often less good, but at ten percent the cost of the mainframe it doesn't have to be to be tempting to a lot of users. This is an attempt to ensure that anything you can do with cluster of Xen machines, you can do with a mainframe (the converse is not true).
I am TheRaven on Soylent News
Contrary to the impression left by the misleading title, this is NOT Windows running on a mainframe. It is Windows running on a blade in a blade center attached to and managed by the mainframe. Using a Windows (or Linux, or AIX) box to perform analytics on mainframe data is not new. What is new is the methods for getting the data from the mainframe, and the fact that the whole thing is managed by the mainframe. And in the mainframe sector, management is huge.
I manage the storage for a mainframe environment for a large retailer. Our busy season is pretty much now through the end of January. Not only are we in a 'holiday freeze', but we procure additional resources from IBM during this season and then IBM takes them back when we no longer need the resources at the end of the holiday season.
On the other hand, my coworkers who work in open systems, install quite a bit new hardware every August/September in preparation for the holiday rush, and then it sits idle come February/March.
Why it may seem expensive, it is quite efficient to be able to have X number of CPUs in use, and Y CPUs physically installed but not leased, so that if we get crushed at 8am on 'black friday', the admins or our management software can enable those CPUs based upon load. We of course get charged for it. If our load doesn't need the CPU cycles, then we don't enable them and subsequently don't get charged for them.
Are you kidding? This is the greatest job security coup of all time! World wide IT departments will have to start hiring around the clock. Think of all that buggy, crashing, virus invected software that will constantly require fixing. This will make all that Y2k hype seem like a walk in the park LOL!!!
Lots of companies use mainframes still. For tasks that require high availability and high I/O, mainframes are your best bet. While you can run a web server on a mainframe, it isn't utilizing the advantages. Running your financial systems where you get tens or hundreds of thousands concurrent users making transcactions is where mainframes have no equal. Also remember it isn't always an either/or situation. A company can use a farm of web servers to handle the front end while the backend processing is handled by a mainframe.
Well, there's spam egg sausage and spam, that's not got much spam in it.
I really see this as clusters taking over a niche where there was no real competition. People would use either mainframes or some kind of distributed solution for that, but that was mostly because it was the best alternative. "If all you have is a hammer, you should treat everything as a nail" and all that.
Of course vendor will fight back, since it will cost them profits. But I simply can't see clusters taking over the real mainframe market.
morcego
The ZBX is designed to replace Racks of X86 servers. The shops that want this either already have a big backend Mainframe - for DB2/ADABAS/IMS with midrange Window/Unix servers or they process everything on the mainframe and FTP down to Windows/Unix boxes (last place I worked FTPed Terrabytes every night down from the mainframe to servers).
The ZBX has a high speed bus connection between the Mainframe (Z196/Z114). This speeds up the network lag for large MQSeries systems, FTPs, etc. Also the ZBX is managed/upgraded by IBM Customer Engineers so the firmware will be IBM supported. They also integrate the ZBX into the Hardware Management Console to have a single point of control. I believe the ZBX can also take advantage of Server Time Protocol so the mainframe can be used as a time source for all the ZBX blade servers.
Shops that currently do not have a mainframe probably won't be interested in the ZBX and the ability to run Windows on it. The ZBX has been out for a year or so but was Linux only until this announcement.
How about that, Microsoft almost catching up with Linux in yet another category. How long has it been ince Lunux ran on mainframes? Quite a while, one of the ten fastest computers in the world runs on Linux (keep working on it, MS). I keep thinking of BSoDs, do you know how damned long it takes to boot a mainframe? Will they have to restart the mainframe on Patch Tuesday every month? Reboot it when its antivirus needs new definitions, or Adobe updates Flash?
Run for your lives, indeed!
Free Martian Whores!
Windows does not run on the mainframe (z/Architecture engine). Windows runs on an Intel blade in a blade center connected to the mainframe with some high-speed links and is managed by the mainframe. The mainframe is still running z/OS, and will have the same performance and reliability characteristics it always had.
Microsoft should know the principle of network externals better than anyone. In computing you often can't dethrone the status quo with a better product, much less an inferior one (and I'm going to guess the Windows solution is inferior in this case if, for no other reason, lack of access to the source code). It is this principle that keeps Windows alive on the desktop in the face of better solutions - and it is what allowed IE to hang on as long as it did.
Microsoft would be better served trying to make some presence on the phone market before it is too late. iPhone and Android are already entrenched to the point that where phones a traditional market Microsoft would be utterly doomed. But they get a saving grace in that phone contracts and devices tend to rotate about once every 2 years. That rapid rotation might give them a chance, otherwise they are shackled to their desktop market - a market that is now just as irrelevant to the future as the mainframe market that IBM lorded over the computing world with back in the 1980's, until Microsoft themselves dethroned Big Blue.
This doomed foray into big iron isn't any more likely to succeed now than it was in the 1980's. IBM has most of the share and none of the players in the field want to have anything to do with Microsoft. These machines are being used by engineers who want total control over the hardware they own and expect nothing less - which is why Linux is the dominant OS and the other major OS'es are open source. I doubt Microsoft really even understands the market they are trying to enter. On the whole its a waste of their time and resources.
The mainframe is still running z/OS, and will have the same performance and reliability characteristics it always had.
Having worked on the mainframe platform for the last 20 years, I can tell you that mainframe reliability isn't what it used to be.
'The tyrant will always find pretext for his tyranny.' - Aesop's Fables
an HP 3000 is a "minicomputer" (in the jargon of the 70s to early 90s) that runs MPE/IX (somewhat unix-like). You just worked at a place where silly people called any server that wasn't a PC a "mainframe". I see that often in my job, manufacturing plants still running AS/400 and calling it a "mainframe", when it's just what was called at the time a "mid-range computer". I've seen Unix microcomputers and Vax called mainframes too, even DEC tried to advertise its big Vax 10000 as a "Vax mainframe", but of course it didn't have a mainframe architecture, had the same architecture as any other minicomputer.
Only one of the last 6 companies I worked for DIDN'T have a mainframe.
Not only does my current company still have a mainframe- we're doing a major software upgrade on it next year.
The mainframe never died.
Mainframe computers were designed around the idea of doing a large volume of repetitive transactions... and mainframes do that very well. If that's what you need done, a mainframe is actually quite a good choice if you can deal with the operational and maintenance costs.
OCO is Loco
And like always, today's mainframe will be on your desktop in 20 years.
Well, yesterday's mainframe is on my laptop today. I just installed MVS 3.8j running under Hercules emulation on a MBP this weekend. Seeing the MVS console running in 3270 emulation (via TCP) again on my laptop screen after all these years was really something. Not useful, mind you, but damn cool.
My favorite job of all times was being a mainframe computer operator in the 80s. Once the suits went home, we'd lock the computer room doors, go pass around a fat one in the decollating room, queue up all of the evening's batch jobs and then watch the tapes spin and feel the floor shake from 40 washing-machine sized hard drives hammering away at the work.
We had a laser printer the size of a small truck that could empty a box of 4400 pages of paper in about six minutes or less.
Oh, and we restarted the system about twice a year - it was so reliable, the sysadmins could halt the processor, apply system patches directly in RAM and pick up exactly where they left off.
The company I worked for leased the whole setup for about $85,000 a month (including an on-site IBM engineer who was there from 8-5 M-F and could arrive on premises within 30 minutes during off hours).
Now I'm running it on a Mac for free, and it's about a thousand times more powerful than that room full of equipment.
Amazing.
Ask Me About... The 80's!