Rumors are flying within IBM this week that the z Systems (mainframe) division is up for sale with the most likely buyer being Hitachi. It’s all a big secret, of course, because IBM management doesn’t tell IBM workers anything, but the idea is certainly consistent with Big Blue’s determination to cut costs and raise cash for more share buybacks. And the murmurs are simply too loud to be meaningless. Think of this news in terms of a statement made last week by an IBM senior executive: “In a world of Cloud Computing, it does not matter what equipment or whose hardware the cloud runs on. We are a Cloud company…”
This move by IBM would not surprise me in a bit. It is my guess IBM wants someone else to make and support the hardware. They’ll be happy to sell time sharing services, AKA cloud services. They’ll be happy to let someone else sell and maintain systems.
When this deal closes nothing much will seem to change for awhile, but in the mid term Hitachi (or some other acquirer) is very likely to port the acquired software (MVS, IMS, CICS, VM/CMS, compilers, etc.) to an Intel-based platform.
A multi-core Intel processor is as powerful as a Power chip in a mainframe, and vastly more powerful than the mainframes IBM made in the 1990’s. Commodity chips can do mainframe computing. By porting the OS and software to Intel, Hitachi can free this market from proprietary hardware technology while still maintaining control of the software. The trick will be to host it on a platform that can distribute the load across lots of processors and provide the desired reliability and redundancy. Imagine an Intel based hypervisor with its good SDN (software defined networking) and SDS (software defined storage) now able to support mainframe services.
Instead of turning the mainframe into a hypervisor to run linux virtual machines, Hitachi would create mainframe VMs that could run on an Intel based hypervisor.
Once this is done, they could port mainframe applications to much cheaper computers, or to a low cost Intel cloud. If Hitachi can’t run these apps on Intel for a tenth of the current cost, they are not trying very hard. Customers would have the ability to host their mainframe applications on the same platform as their non-mainframe apps, and host them anywhere they want.
IBM should have done this years ago. If only they hadn’t sold to Lenovo their Intel-based server business.
This is what happens when you stop doing development, improving products, etc. The biggest reason IBM’s system sales are floundering is they’ve neglected the products and their customers. Hopefully this rumor is true and someone else will start doing what IBM should have done…
at this point, really, it’s a replacement market. and dwindling. IBM has sold its fab, offed its staff, sent support overseas… wherever, it shifts and changes, probably on whichever island is above water every third Tuesday by now.
so now they want to sell the hardware manufacturing, but keep the ancient software with yellowing notes flapping in the breeze in empty cubes that hold the internal secrets.
that’s like trying to score every semi-working VAX-8000 to keep an application alive.
if I was Mr. CTO, looking over a floor of red or dark grey capital investment, I’d be very worried.
John,
Great bombastic start,,, where is the first detail?
Go back and read the last 10+ years of Cringely’s work.
You mean like when he predicted IBM would lay off 25% of the population last year?
I have NEVER read such rubbish – and I probably know more about the mainframe and IBM than anyone else in the World with over 50 years experience. The mainframe, related software and maintenance make the majority of its profits – the other businesses they sold made NO money.
I disagree with you. First, I worked at that company over 50 years ago and knew many of the major players. Second, nearly every business that was sold off, has been profitable in someone else’s hands. The blundering started in the 70s when there was an attempt to produce FS (Future Systems) but the H/W required was not within reach, Next there was the attempt to get the “entire world” on the large OS (OS/VS2 nee MVS) and eventually kill the lower-end OSes. This did not fly well in the WTC countries, where a lot of the businesses were still “small to mid-size” and unable to afford the infrastructure to run the large systems. One of the last straws was the advent of the PC. Management was advised to develop the business totally “in-house”, but its attitude back in the late 70s was PCs were “kids toys” and for hobbyists, forgetting that the successful 1400 series and the early 360s had less processing power that the chips in those early PCs. In addition, they still had the S/W developers who worked on those early systems (some of whom contributed in-house to the PC S/W). Management still had its eye on moving customers to VS/2 and its successors because it required major H/W to run it and that was big bucks – a case of short-sightedness. BTW, keep in mind that back in the mid-to-late 70s, Unix was starting to pick up steam – while not totally ignored, it was not taken as seriously as it should.
No – others have not done a good job of profiting from businesses IBM sold, with the possible exception of Lenovo, which absorbed the IBM PC business rather than taking it over as a new line of business. Whatever criticisms might be leveled at the company, it is quite good at determining when it can no longer compete with commoditizers and should leave a business and apply the money elsewhere.
The problem is that there is no place to put the money except toward share buy-backs. Eventually the company will cease to exist. There will be no business left where they can compete.
Think about what moving zOS to tinker toys would do to the likes of BMC, CA, SAS, Compuware, etc etc. Not real sure the DOJ will allow a sale with the thought of getting rid of zOS and porting given the impact to other companies. Then again, not sure the DOJ would realize the impact. As stated below though, doesn’t make a lot of sense to bother to port. One could rewrite apps on Linux cheaper and faster.
Just saw this today on Ars Technica. It appears that IBM isn’t alone in eating its seed-corn and creative bookkeeping.
http://arstechnica.com/information-technology/2016/06/oracle-boosts-cloud-sales-but-at-expense-of-java-and-everything-else/
The book “Putt’s Law and the Successful Technocrat” by Archibald Putt is an excellent read for anyone interested in understanding why tech giants fail from the inside out. The people that actually understand the technology are not corporate ladder climbers, when they leave the roof falls in.
mainframe emulators are available that allow z/os to run on intel computers.
see https://en.wikipedia.org/wiki/Hercules_(emulator)
IBM MAINFRAMES ARE TOO EXPENSIVE AND INTEL/ MICROSOFT/ APACHE SERVERS ARE
TOO CHEAP!!!
The IBM brand does not have the prestige, power and positioning it once had!
My customers now prefer iNTEL, LINUX, Microsoft or Apple rather than IBM!
I have been using IBM and INTEL/ microsoft computers for 30 years in my business
Computers are just too CHEAP to sell IBM z systems to be used as linux servers!!!
ONLY COMPANIES WITH LARGE IBM SOFTWARE COLLECTIONS WOULD EVEN
CONSIDER BUYING A Z SYSEM
For legacy customer, IBM z systems are great!!! For the other 99 percent it is just over priced!
Back in the 1980’s 1990’s, I was an ibm mainframe dealer and custom PC mfg,
In my service/ consulting business I have installed, maintained, programed, repaired the following ibm
hardware: IBM 360, IBM 370, IBM 400, IBM system 3, 34, 390, z series machines.
I LOVED PROGRAMING AND REPAIRING IBM HARDWARE!!!
When I used IBM hardware I had to maintain an office, wharehouse and staff.
Now I use 12 laptops and desktop servers to run my business, from my apartment!
ITS A DIFFRENT WORLD!!
YES IBM IS REALLY JUST A CONGLOMERATE OF SOFTWARE COMPANIES!!!
Will cloud computing be successful for IBM?????
This sounds like a return to the 1970s when ibm owned the service bureau corp and gave it
to control data, to settle an antitrust suite.
You knew times were changing when IBM and Jersey tab stopped making punch cards in the 1980’s
The problem is profit margins. The Mainframe business itself could probably be considerably cheaper but the companies have come to like the profit margins they get.
Kind of like Cisco – cheap to make gear sold at insane prices; their $200k gear does not really cost $200k, more likely $10k at most if not closer to $1k; and the services provided are not really that expensive either. So they’ve picked up quite a hug profit margin.
IBM’s Mainframe business is likely not that different from Cisco in that respect. Sure they’ll charge a few million for a mainframe, and give support, etc; but their up-charge to do so has a pretty sweet margin for them.
Anyone buying the Mainframe business from IBM is not likely going to want to give that up. Porting the functionality off Power means that a different vendor could potentially be used, and therefore the margins would have to come down if they want to keep the business. So I wouldn’t expect any buyer to change something that would affect their profit margins on their newly acquired business.
It’s looking more and more as if IBM is just a vehicle to generate a nice fat retirement account for the people running the company – everything worth anything is being sold off. Once upon a time it used to be said that the most valuable asset in a company was its employees, these days they are treated as a liability – there’s more money to be made in scrapping the mainframes for the gold in the components.
True, here’s one. Should it be true, this would be the second huge mistake they did after leaving people at home to keep stock profits up. Shame. Seeing IBM selling what has been its core business for decades again for a mere money reason it’s something that left me at a loss of words. Past and gone are the times when Gerstner ruled. Where I live I heard the so called cloud division was already failed even before it was born. I am so displeased for my ex colleagues that will have to face a future totally uncertain only for some wanted to make rich people richer. Shame, again. Will my colleagues end up repairing low tech Chinese black boxes at the local gas station? No idea but I really see a decline of something, pride, knowledge, uniqueness, that has always been priceless until someone decided it had such a meagre value, money. Today you have it, tomorrow you don’t. Soon however top managers will eat their own cr@ppy food, that’s for sure. Best of luck guys, but it won’t be enough, I’m afraid.
Looks like eventually IBM will have sold everything that they own to buy back shares and a few shareholders will own a company with no employees, no income, and no assets except the name IBM, which they’ll probably sell to some Chinese company that’ll slap it on cheap junk for a few years until all potential customers have caught on.
So IBM is the next Trump?
Bang on!
I love how you cucks desperately hate on the Trump.
No, Trump wants to bring back jobs to the US, not ship them offshore as this deal would do if it happens.
More likely the next Hillary, who is already in bed with the likes of Wall Street and Corporate “greeders” who use buybacks in an attempt to keep the share price up. All those high-tech workers made up a large part of the middle class, but doling out H1Bs like candy, shipping those high-tech jobs overseas helped to decimate the the middle class. While one may not like Trump’s brashness and “opens mouth before engaging brain”, he is not the cheat and liar like the Clintons. Nor would he sell out his own mother like those two. Too bad the media is just as dishonest as them.
Re: “Corporate “greeders” who use buybacks in an attempt to keep the share price up.” If the owner of a small business, or his delegated advisor, decided to close down his business, he would simply buy out any other owners, if there were any. That would have the affect of increasing the eps ratio just as reducing costs by firing people would. If the market perceived the company was so valuable that it deserves to be kept in business, someone else would simply buy out the current owner who just wants to get out. One could argue that all business is based on “greed”, depending on how you define it. The political comments are a bit off topic, but since they were brought up in this thread, I should mention that I do agree that Trump is best candidate, since only he remembers that America used to be great, among all nations, providing a suitable goal for the Federal government.
Even if they sell all their products, IBM will still have thousands of patents. I see them turning into a patent troll once they give up on making things.
Is MVS et al. written in Assembler (BAL)? How are they going to port that to Intel? Or has it been converted to C?
If they haven’t ported it to C already they could do it the same way they did it with Pacific (aka S/38 aka “System/3 Successor” aka STS aka FS “Future System”) or the S/38-AS/400 High Level Machine Interface which could be part of the sale. They could do it via an FPGA implementation of the old IBM 7436 Hybrid Technology Workstation as a iCORE attached co-processor.
Very knowledgeable reply. Nice to have specifics.
Yeaaah…no. Too many hardware dependencies, too much “clever” assembler code, too few people who understand it. Would be more practical to rewrite from scratch (and that’s not even close to practical).
IBM already owns a version of z/OS that runs on Intel Hardware. It’s called RD&T, and they’ve been positioning it as a developer tool. Basically Hitachi or whomever buys up IBM’s IP and starts working hard to make RD&T ready to run across a cluster of Intel commodity hardware, and the IBM hardware business disappears. Everybody wins except IBM hardware engineers.
Re: “Everybody wins except IBM hardware engineers.” Shouldn’t that include the software engineers and IBM employees in general, since that entire business was taken over by Hitachi?
z/OS is written in PL/S, an inhouse variant of PL/I, with some assembly language parts. No doubt it could be ported over to C with some effort.
Why port it to C? Why not just write a compiler like they’ve done each time they switched chips.
They have the PL/S (PLAS) compiler and could have marketed it, but never wanted to impact their PL/I sales – which, today are probably non-existent. Two things where IBM could have done better was making PL/S available (better language than C) and pushing VM (zVM) harder back then. But again, it was believed that they weren’t important to the companies business and the the large OS (with PL/I and COBOL) was where the money was to be made. Short term thinking. Keep in mind, VM and Unix sprang from the same Multics project (ATT was initially involved).
I can’t tell with whom you’re agreeing. Should it now be ported to C or not?
ibm operating systems are written in assembly language NOT pl1.
Very few people ever used pl1, ibm programming was done in cobol, rpg or assembler!!
Ibm assembler is a very powerful macro assembler(best assembler ever made)
I thought there is a cross system variant of PL/S. Seems more likely a port to it. Plus, IBM owns the T3 tech that changed Intel processors microcode to place z on Intel native or nearly so. Between that and the RD/z that you can run on Amazon cloud, there is little doubt it is already a reality. Additionally, there is the DB2 Pure Scale which has Coupling facility code running on Intel.
PL/X not PL/S
Already have a zOS simulation sw running on PCs, so moving to C won’t be hard if all they do is simulate it. But, I agree with everyone, it could be written in C. What a mess it will truly be though.
As of what I’ve learned from IBM people, lots of functionality in today’s z/OS is already written in low-level C
ibm operating systems are written in assembly language NOT pl1.
Very few people ever used pl1, ibm programming was done in cobol, rpg or assembler!!
Ibm assembler is a very powerful macro assembler(best assembler ever made)
ibm never used c for systems programming!!!!!!!
EVOLUTION NOT REVOLUTION!!!!
ZOS IS 75% 360 OS MVT(1974 ERA), ADD ON CODE TO THE ORIGINAL IBM 360
CODE BASE OF 4 MILLION LINES OF ASSEMBLY LANGUAGE CODE(BAL)
this may sound naïve in terms of business sense but I just have this feeling that if IBM had stayed in the personal computer market they would be in better shape.
Now I know they probably didn’t have the profit margins in order to keep sustaining the business but I think they should’ve accepted those smaller margins so that they could keep their name in the hearts and minds of people especially businesspeople where it would be an easier segway into selling the services that they currently sell.
If your name is not out there in business you become easy to forget
what IBM should have done is… accepted lesser margins, tightened the oversight chain, cut all the proprietary interconnects to off the shelf stuff, concentrated on value as well as power, and tripled their customer base. or more. kept some rogue shops like Boca Raton and accepted the innovation as long as it could communicate.
but no, total control from Armonk, all wisdom comes down, all bonus flows up.
Apple has the profit margins for premium hardware and premium software
It’s those very profit margins that keeps businesses, and cost conscious consumers away from Apple products.
As a developer, I value a premium experience in my laptops. I used to buy Thinkpads because of the superior keyboards and how well they were put together. The Dell machines of the time had that plastic-on-plastic squeak as they flexed – Thinkpads had magnesium chassis and didn’t flex. IBM Personal Systems Group changed the keyboard shortly before the sale to Lenovo, and that’s when I started buying Macbook Pro laptops.
The keyboards weren’t as good, and it took some time to get used to OS X, but the build quality was excellent (solid aluminum chassis) and it had great style. But over time, Apple gradually started removing features and made it far less “Pro” than it used to be. Internal optical drive? Gone. Then went the internal Ethernet port and the replaceable battery, and they went to a “Chiclet” style keyboard. All in their quest for the slimmest laptop possible — something I don’t care about. I want a machine that I can get work done on, so now I’m on a Microsoft Surface Pro. The clip-on keyboard is nothing like the original Thinkpad ones, but it’s tolerable for when I’m away from my external full-size.
I agree with you. However, management was short-sighted back then and, IMO, even moreso now.
I love the picture. Someday, we’ll all be working on that great big Mainframe in the Cloud.
you DO realize that Cloud is the same thing as Timeshare, right? with rented storage.
Possibly, you might be missing some or many Cloud potentials. Alluding to Cloud as a later day CMS.
Actually, he is not far off (except, maybe older terminology). When you stop and think about it, the applications were on a server (another CMS “machine”) and your “data requests” were processed by the “client” support on your A-disk (or via another server acting as an intermediate). When you make a database request, the DB is accessible on a server (cloud). You develop S/W, the compiler is on the server (cloud). Nothing is really ever new in computer technology, just a new buzzword to say the same thing with maybe a slightly different twist,
You’re right SWSCHARD!
It’s back to the future with Timesharing (aka Cloud).
50 years later and just renaming a great service!
Here is a good youtube video where our company president covers it well.
https://www.youtube.com/watch?v=_Fgsam53mM8
https://www.youtube.com/channel/UCkstNWqQoJhBqZmRzbvR_ow
Just to be clear, the MF has always been a cloud. So, IBM has always been a Cloud company in the MF space. I guess this goes to the point already stated…..playing on customer’s lack of computing knowledge – not to mention Wall Street’s, who think Cloud is something new – it’s not!
While I agree with you that the “cloud” is not something new, I wouldn’t call the “mainframe” the “cloud”. Prior to the 1969 unbundling (the consent decree IBM signed with the DOJ), it was running a “cloud” in its data centers (Service Bureau Corp.). Customers did not have to lease (or buy) a mainframe to run their payroll, etc. They took the work to an IBM SBC data center, paid for a time slot and ran the job paying only for the time used and the machine processing power. Today, they can accomplish the same thing via the internet instead of carrying a box of cards to a data center. After IBM was forced to unbundle S/W from H/W, IBM also had to divest itself of the SBC (as I recall, I believe they were sold to CDC). Actually, there was 2 SBC unbundlings (to speak), 1956 and 1969, Trying to remember the name of the Service Bureau Corp., I came across this article: http://mainframe.typepad.com/blog/2011/09/did-the-us-government-kill-cloud-computing-over-50-years-ago.html .
“We are a Cloud company…”
Is this the closest that any corporation has ever come to admitting in broad public (as opposed to behind some number of closed doors) that their business model is predicated on the stupidity of their customers?
Thanks for telling like it is! 🙂
Vapourware!
This is the company that gave us the 360 series. Wow. Just, wow.
40 years ago?… It was a ‘wow’, but now a quite old wow.
50 years ago! You lost a decade…
Don’t forget the Correcting Selectric III and the Wheelwriter and all the typewriter repairmen (such as I) who found themselves on the street due to “restructuring”
I’m not so sure Hitachi or anybody is going to find it worthwhile to port z/OS or any of the applications to Intel. The software is tied pretty tightly to the underlying hardware capabilities. It would probably be a lot easier to just transplant the good bits into Linux. Sure the zSeries cpu is nothing special, but there’s no Intel box that can push i/o rates like the zSeries channel subsystem. Sure my raid card can do 200MB/sec, but a zbox can have 200+ channels that can do 100MB/sec. No, I expect that z/OS will just stay a niche system.
imagine a Beowulf cluster of Intel clusters running I/O instances with z/OS….
Exactly! The mainframe is an injection beast that keeps its’ CPUs running full tilt. It is all about throughput. Me? 44 yrs m/f, MFT/360, MVS, VM, zVM and zOS. Hone/IBMLink global support. IBM RA’d 2004. Bounced back better than ever, still going strong. Sad to see the decline of IBM, but predictable.
While the z enjoys fantastic IO throughput on the z/OS side using the SAPs the Linux on z side is totally different where the SAPs are not used by Linux.
Exactly Art! Then again, tinker toy lovers have no concept of I/O channels and don’t understand it is the main reason, if not the only, that they have to run their baby boxes around 40% to get any type of performance. Don’t get me started lol Given the MF’s availability and near instant DR with minimal HW outlay it will be a stupid move. What is going on is those in executive mgmt. today only understand tinker toy boxes, thus everyone gravitates to what they know. Little do they know their tinker toys are easily hacked and their availability – even with redundancy – is horrid. We’ll see how many millions it costs a company when their tinker toy is hacked.
Yeah, that’s why they bought PSI (Platform Solutions, Inc) and shoved them in a closet. z/OS is just to complicated to run on Intel chipsets. It’s already been done. Granted the Itanium (Itanic 🙂 ) wasn’t the greatest chip Intel ever made, but you certainly could fab one if you desired. Further, z/OS is not tied to the hardware anyway, maybe the microcode, but not the hardware.
Rather reminds me of the Stephen King short story: Survivor Type. See description, here: https://en.wikipedia.org/wiki/Survivor_Type
Hard to believe….
Can z/OS architecture be ported to Intel? Yes, but the cost would be very high and performance would be very poor. In other words, z/OS would behave like windows or linux…
Another question: How will be handled channel, parallel sysplex and other z/OS concepts?
How do you know performance would be poor?
I seem to remember not too long ago when industry pundits were convinced that PCs where just toys.
Z/OS and the hardware it is on are very dedicated configurations for the task. Sure, they could port the stuff to a variant of the Linux Kernel and take advantage of Open Source, but even then too much wouldn’t be optimized for the kinds of stuff that actually runs on z/OS nor for the hardware. So it will naturally be less optimized and therefore less performant in many critical areas that are outside the domain of where software that normally runs on any other system needs to be optimized, or in ways that are only optimal for the use-cases of that hardware and/or software solution set.
FYI – Linux is already on the mainframes already – same hardware, so it doing the port could be a solution, but only if performance matched, which is yet to be seen since Linux’s optimizations or even those in the main frame port are not likely to be the required optimizations for what z/OS does.
I know Linux runs in mainframes. It’s been there for a long time. Since year 2000. Linux for system Z takes advantage of many features that only exists in Z systems.
As Gene used to say….. FUD
Gene?
Gene Amdahl
They are just toys, especially up again a MF.
If you knew a bit about z/OS and Intel architecture you would know that…
But here is a short explanation:
Mainframe have special processors dedicated to specific tasks. To handle I/O mainframe have channels that use dedicated processors and that allows main processor (CPU) to run other services while I/O is done.
Intel doesn’t have that concept so, while your operating system is doing I/O all other tasks have to wait until CPU is free.
There are many other things too but it would make explanation too long.
And last but not least mainframe processor clock works at 5.5 Gigahertz while Intel processors works at 3.5 Gigahertz.
Intel has had this ability for a long time… DMA.
So does that mean the only advantage of a MF over a PC is the speed, 3.5 vs 5.5 GHz? Or do the so-called “channels” do other things beside I/O, providing a significant advantage.?
z/OS porting to Intel (Windows and Linux) has already been done (Hercules, zPCR).
That is no problem. Performance, reliability and recoverability are the key. Without that noboby in his right mind would get away from zSystems.
Some observations. Running z/OS on Intel is of course very do-able refer to: Hercules, zPDT, Rational Development & Test for z. So functionally (on a small scale) – no problem. Developing solutions for high-end customers: supporting massive OLTP and corporate systems or record, particularly with Parallel Sysplex – this would need development to run on Intel. But most things are just a small matter of money and programming …one’s definition of ‘small’ may vary 😉 In other words, while hard, it’s do-able.
However, whether z continues to run on z processors (not Power, Bob!), or Intel or whatever – that’s not that interesting.
What is interesting is that if this rumor is true, then IBM will finally be slaughtering the cash cow that it’s been milking for so long (and has played a key role in providing cash to fund development in other parts of the organization). To pick an analogy out of the air – this would be like selling your heart and by the power of magical thinking, hope that you can get by and actually improve things now you’ve got rid of that pesky thing that was really just ‘holding you back’… this may sound silly, but that’s effectively what will happen… it won’t be pretty for the broader company. If the rumour becomes fact, this perfectly illustrates the level of greed and delusion alive and well at IBM. Saying all that, bring it on – the Mainframe hardware and z/OS are the best things IBM have ever done, they still power most of civilization and under the stewardship of a new owner I fully expect they will thrive (as opposed to the current neglect and abuse that’s been all too prevalent under IBM’s watch).
Exactly what I was wondering…how can IBM sell its cash cow ? How can this make business sense ? Without mainframe we are only left with Power and Storage, so I wouldn’t see them holding for long…and then the whole hardware business would be gone. I’m not a technical person, but can IBM do that fancy stuff with IA (Watson and the like) on Intel-based HW manufactured by somebody else ?
Agreed!
Why would IBM give up the cash cow and the very best thing IBM ever created?
This kind of thinking would surely put a nail in IBM’s coffin.
Selling services is great, but not comfortable without control of the infrastructure those services require.
And: “The trick will be to host it on a platform that can distribute the load across lots of processors and provide the desired reliability and redundancy.” to re-quote the article.
This supports the rumor I heard last month that the entire mainframe DB2 staff had been axed. Something like 192 folks in Russia where it had been moved some years ago.
This rumor has no basis in fact. DB2 for z/OS development continues at SVL in San Jose.
Nahhhh. Never. Well, probably not. Although I suppose it’s possible. But what about the Power systems line? And if they did sell these two, would they knock out the TSS (maintenance) division too?
Ok, so the author doesn’t know midrange from a mainframe. If the rumor is true his conclusions are still way off base.
The OS and program products cannot be ported. They are architecturally dependent. Unlike the horsepoop being peddled by various companies the IBM mainframe and software that run on it are designed to run together and have been since the very beginning. There is absolutely no way to port the OS or the program products or even most of the applications. You can quote me on that. Not going to happen, wouldn’t be prudent. The OS can not be rewritten in C. Can not.
Now what they have done long ago is emulate the hardware. And they have done that in several interesting ways with several different implementation twists. There is even a new company claiming to move mainframe workloads onto toy computers. There was the PC/370, the P/390, FlexES, MP 3000, there’s RDz and there’s good old Hercules. People who know understand it’s easier to emulate the hardware than the OS and applications. The way to solve this would be for a company like Intel or Fujistsu or whoever has enough cash and a brass pair to create commodity server boxes that have z compatible chips in them so they wouldn’t have to emulate. But that doesn’t get rid of the scalability (z up, Intel out) issues, nor the I/O issues, nor the management issues.
What you don’t understand is all the PCs in the world hot glued together can not run the workload of a mainframe. A mainframe is bigger than the sum of its parts. It’s not just about clock speeds. The latest z models win soundly there too but that’s far from all. It’s hardware and software optimized over many decades to run together symbiotically, it’s I/O devices and channels designed to take virtually all of the I/O processing away from the mainframe and move it out to external devices and a whole lot more including the ability to run and administer and workload manage humongous workloads. It will certainly be possible to split application by application off onto commodity crapware. But then they’ll be right back to all the problems that have stopped PCs from taking work from the mainframe. Centralized management, reporting, and control goes away and you’re back to a huge data center full small PC clusters hosting one logical application and no way for those applications to share data across the enterprise (which they already do more or less seamlessly) no automatic failover, no load balancing (I mean workload, not I/O).
Ain’t gonna happen folks. There’s a whole lot of money been spent on this over the last 50 years. If they could get away from it and still get their work done they would have done it. The mainframe solves a ton of problems, stays up 5 nines and gets more work done than a datacenter full of anything else. They’re worth price paid. If IBM does sell the z line the smartest thing the buyer could ever do is not change anything.
It took me a while to figure out this was satire. Good one!
Well said. Throughput is key.
Mainframes have already been replaced by commodity hardware in many cases.
Mainframes are not a growth market.
> Mainframes have already been replaced by commodity hardware in many cases.
Nonsense. The commodity hardware/software solution always fails and instead of replacing anything the data center just runs more of a mixed bag of doggy-do. In the few situations where the mainframe has been removed after years or decades of trying to get it out the door the applications still don’t work, all the qualified staff is gone, and the remaining workers are clueless. In the end very little is delivered and the mainframe hasn’t been replaced in any meaningful sense of the word.
> Mainframes are not a growth market.
So what?
As some great venture capitalist said, if your not growing your dead. You just may not know it yet.
I’d point out Google, Facebook, and much earlier, eBay. They all ran on on commodity hardware, in some cases significantly below the current state of the PC art (I remember Google running on 486’s, when Pentiums were current. 486’s were much cheaper, and therefore more cost effective). They made the PCs spread the load across themselves in the matrix, so that you could lose a PC, or more, and the system would continue processing. In that respect, it was like a Tandem computer, where you could remove pieces of the computer while it was operating, and it would continue working. Also, a matrix of PCs can handle a much larger load quickly. Example: Google’s real-time fill-in, where it guesses what you are looking for, while you are typing in in, and note, it is doing that on a world-wide basis.
Spreading the load out allows commodity hardware to handle very large loads, without needing the channel hardware that a mainframe has, and it occurs to me that another example is IBM’s Watson, which isn’t running on a z Mainframe.
Hey IBM Guy,,, Outstanding posting!!!! Nice opening zinger between the Eyes of Cringley, you nailed his weakness,,, he does not know IBM Main Frames
Right on IBMguy. If the changes they make come true then it would be a huge step backwards in thruput. Hope you don’t mind waiting longer for all that instant processing that happens today. Also how secure will your data be and how fast will the Cryptographic operations take without the mainframe. Seriously when was the last time you heard of a outside virus getting into a IBM Mainframe? Are you going to trust the cloud running on PC’s?
Isn’t Amazon’s cloud already running on PCs?? And Microsofts, and Oracles, and…
>There is absolutely no way to port the OS or the program products or even most of the applications.
Turing weeps.
IBM might be wanting to flog off large parts of its legacy business, but whoever buys it sure as anything won’t be porting code. Its all assembler, HLASM, PL/I, PL/X, horrible crap known only by the people at the target RA age (50+).
IBM has not axed their entire DB2 z/OS organization. It has been hit hard by layoffs, retirements (not to mention the yet to be seen impact of the new colocation/no-work-from-home policy), but they are still plodding along, albeit at an ever decreasing rate of productivity and innovation.
The way I see it, internally we see the message that mainframe software/hardware is fundamentally incompatible to the future vision of the company. Even though it brings in much money, it is not GROWTH, any investment on mainframe is investment taken away from their cloud first strategy, which is not tolerated. And so the only and ultimate solution needs to be drastic – to remove the distraction and divest that part of the business.
Exactly correct. But as one of those 50+ RA candidates, I sure hope this happens and Hitachi invests in z/OS
Agreed. As a former IBMer and z/OS veteran, I would be quite happy to entertain consulting gigs for Hitachi and/or their business partners if this rumor came to pass. As it sits now, if an IBM recruiter called me in search of z/OS skill in support of an IBM owned contract, I would laugh them off the phone.
Also, DB2 might not be part of the deal. It is part of Analytics not Systems – and runs on many other platforms besides Z. There was a lot of speculation after the recent reorg that separated DB2 from the rest of the Z software products that a Systems Group sale was in the works but IBM would keep DB2.
Given the importance of DB2 to Z customers, any deal would have to address this issue.
Possibly IBM and Hitachi would agree to work together to keep DB2 alive on Z – and DB2 Z is tightly bound to Z hardware. There are many Z hardware and z/OS system software features and functions implemented specifically for DB2. How would this work ( especially going forward ) with DB2 remaining with IBM?
From the Hitachi view, the z/OS business line would be is highly dependent on DB2 and customers would absolutely need to feel secure about this issue. Virtually every Z mainframe customer has DB2.
This way IBM would get to keep some cash-cow revenue with Big Iron DB2 and not have to go cold turkey into the brave new world.
My last two customers ditched their Mainframes. They looked at our bill. They could go out, buy replacement software, buy a mid-range “server” and do the same work for 1/10th of the cost. They did it and were able to recoup their investment in less than a year.
.
While was happening I was getting a LOT of helpful advice from IBM leadership.
.
First they wanted me to sell my customers Linux LPARS on their mainframe. Small problem — their Linux software would not run on it. Even if it did run or if IBM paid to port it, the cost to run it was at least 10x that of doing the same thing on a VMWare cluster. At the insistence of IBM leadership I put together a proposal. The customer looked at our prices and said “are you nuts?”
.
Next leadership wanted me to sell our cloud mainframe service. The cost to move their applications and data to the cloud was quite high. The price to run it in the cloud was less than 20% lower than on their own leased systems. Running it on our cloud wasn’t a good deal and again the customer looked at us, sighed, rolled their eyes, and chastised us for wasting their time.
.
In spite of all this “HELP” I got from leadership I was able to sign a good amount of new business. For that I received $0 in commission, rewarded with an RA and a whole month of severance pay.
.
When your customers are actively looking for ways to get rid of their bills from IBM, that’s not good. There is a reason our revenue has been dropping every quarter for years. As a corporation we’ve become a bunch of idiots. I am glad to be free of the insanity. Trying to talk sense to IBM leadership has been exhausting and I am glad I don’t have to do it anymore.
This is a good move to CLOSE SHOP ! All leftover $$’s go to the Corporate Execs.
This “rumor” does make a lot of sense to me. The stickiness of mainframes are the billions of lines of code that have been written since its inception in 1964 mostly written in higher level languages like Pl/1 and Cobal. To take this code to a low level language like C would make the maintenance of this code almost impossible since its already what we called “spaghetti code” meaning its so unstructured its very difficult to maintain now. I think this is another Cringeley blog attention getter…………………
Marc Andreesen famously said that “software is eating the world”, implying that combining hardware and software in the same company is now a mistake. HP has finally completed that transition, spinning its services business into a combination with CSC. Now it’s IBM’s turn. The resulting Hewlett Packard Enterprise still will collect a good revenue stream for decades to come – nobody is shutting down the stock exchanges that still run their transactional cores on HP NonStop systems and other “scale-up” systems that run on HP-UX. Nor will any bank or airline reservation system be shutting down their Z system mainframes that run 40 year old software until another 40 years have passed. These things are the computer equivalent of the B-52 bombers which were flown by their current pilots’ grandfathers and which will still be in the air when today’s pilots have grandchildren in the Air Force.
HP started transitioning its own mainframe systems from the ill-fated Integrity to mainstream Xeon CPUs more than five years ago. IBM hasn’t started the corresponding transition, and won’t on its own initiative – the z hardware and OS and middleware ecosystem is too deeply interlocked in too many ways. Sending that unit to a new executive team via spinoff is the only solution.
Doesn’t NextGen SABRE run on MySql?
Selling it doesn’t surprise me, but you’ve completely got the porting idea wrong.
The mainframe is far more than just the CPU – backplane, memory and storage bus, and more all are fundamentally different from other architectures. Recompiling for Intel is easy, but the OS is far more than just that, let alone all the 1960’s era apps that still rely on ancient components.
I for one love a good conspiracy theory so I might add that the unbelievably advanced technology incorporated into modern z13 Mainframes is rumored to be of alien origin!
Did it not occur to you that IBM launched the 1st 701 Mainframe in 1952 shortly after the 1947 Roswell UFO incident? The reason IBM chose the 3070 green screen as user interface was so that it would not upset the eyes of the alien overlords from planet Bong …
Seriously, it has been a while since I last read an uninformed article like this:
The IBM z13 microprocessor is the most advanced and fastest commercially available CPU on the market boasting 5GHz, massive caches and a CISC instruction set that bakes all the instructions in actual transistors – and that is just the CPU! In addition, a modern Mainframe boasts hundreds of specialty chips such as FPGAs that handle anything from compression, cryptography and I/O. The z/OS operating system, the Parallel Sysplex shared memory cluster technology and many of the Middle-ware components and applications running on Mainframes have been fine-tuned over decades – quite unlike your latest PC OS that makes that fast Intel chip perform just as well as your dad’s old 386.
Take a look at IBM’s 2015 earnings report and you will see that the Mainframe business is growing double digits. The reason for it is that companies all over the world realize that current IT trends like real-time analytics, mobile computing, Blockchain, cyber security just love a massive, hyper-optimized system that offers 99.999% of availability, the best security track record in the industry and elastic scalability any cloud can only dream of.
Oh, and then there’s the little known fact that our alien overlords from planet Bong would blow up the Earth if IBM were ever to sell its Mainframe business…
Well written sarcasm toward an unknowing Cringley, But the real essence of the article is the detail
of the main frame archetectural features.
In total,,, all of the features working in synchrony, make the Gulfstream “Pursuit Of Perfection” and Lexus variations of “Relentless Pursuit of Perfection” similar to child;s play. FIFTY Plus years of integrating hardware, middleware and software have created a near total system, that few outside of even a small group inside of are even aware of how integrated and closely knit each component is built to work with minimum effort.
I don’t think that Cringely would even be able to talk meaningful about the FPGAS or the Parallel Sysplex cluster technology,, let alone the hundreds of other special application subsystems.
But, on the other hand he was never very much in the high end technical field. Touche Cringely.
Re: “latest PC OS that makes that fast Intel chip perform just as well as your dad’s old 386”, Not necessarily just as well. I remember being surprised “upgrading” from Win98 on a 133 MHz processor to newer versions of Windows running at 1 to 2 GHz, observing Notepad was less responsive. It may be a lack of optimization or just that the OS is trying to do too much.
You are assuming the current IBM execs actually care about great hw/sw. If it brings in cash they’ll do it. It doesn’t have to make sense.
How does that differ from any other business, non-profit, or government tax plan?
@Mainframe Guy: Thank you. At last some arguments that make sense. If this rumour becomes true, then IBM goes completely mad.
Whilst it would be possible to port z/OS and the IBM supplied tool sets surely the real issue is the billions of lines of application code that relies upon the underlying z Architecture.
The concept of shared address spaces with storage protection is totally alien to the x86 architecture and operating systems running on x86 type hardware. Just the simple task of securely getting a transaction into an application through CICS, obtaining data from DB2 and getting the response back to the user, which is a seamless flow of control through the various address spaces in z/OS, requires many full context switches when implemented for the x86 architecture – it is all these context switches that really hurt performance when applications are migrated from z.
Also, whilst applications written in a high level language (assuming you still have the source code) could be recompiled there is still a sizeable amount of assembler for which it would be necessary to provide some sort of Dynamic Binary Translation for – IBM did acquire Transitive a few years ago so the tooling does exist but the performance may be somewhat problematic.
The x86 architecture does support shareable address spaces and it also has key access control to memory
Well eventually I expect IBM will find it too expensive to continue Power Chip development so sooner or later Mainframe systems will stagnant or have to be moved to a new chip architecture – Intel and ARM are probably the only contenders.
Ofcourse if IBM sold it off and then shutdown Power Chip development then it would be someone elses problem. AIX is doomed as well but those folks can move to linux relatively easily.
A large number of Banks and large corporates still use mainframes as their system of record however these systems because they are so slow to update software wise are now causing these companies to lose competitiveness with companies that are much faster at delivering software changes (Google et all).
New fintech startups will threaten the banks in the next 10 years so we maybe at the start of a big move off IBM Mainframes so it might be time for IBM to get the most value by selling.
re; neill turner June 16, 2016 at 1:55 am
>> A large number of Banks and large corporates still use mainframes as their system of record however >> these systems because they are so slow to update software wise are now causing these companies to >>lose competitiveness with companies that are much faster at delivering software changes (Google et all).
From a system software perspective this is absolutely incorrect. Due to the 1,000s of non-mainframe servers large corporates have, updating them is a lesson in painting the Brooklyn Bridge. This is a major reason why they “skip” releases. Where I work mainframe system software updates are on an approximately 3 year cycle. 1 year for moving the updates through environments, 2 years of version/release stability. The company also performs maintenance every 6 months.
From an application software perspective corportates that have customer facing application (e.g. Banks) require stability and reliability over anything else. The one thing that freaks the execs out is for the customer systems to be down as it directly affects the corporate’s reputation and can get the execs’ names in the press – bad for career. The one thing that every non-IT Bank employee wants is accurate customer data – nothing is worse for them then to try figure out bad data while the customer is screaming at them! An expense report being late by a day is one thing, Mobile/Web transactions not working or corrupting data is another. It is not the mainframe per se that are slow to update it is that mainframes are running very critical customer systems that must be accurate. As a result companies have built *processes* to dot the I’s and cross the T’s. I am seeing the same processes being applied to non-mainframe critical systems.
Actually it is safer to make small changes more often than large changes less often. Facebook change there code 10 times a day and no one notices.. That’s why the banks sometimes have very embarrassing outages when they upgrade their systems every 2 years with alot of changes. Unfortunately there is more and more regulative change etc so changing software system once every12 months ain’t going to hack it in the long run.
I believe you are confusing the customer (in this case banks) making changes to their code vs the computer company (in this case IBM) sending out a few periodic updates or an APAR. Regardless, when a computer company turns out a major release, customers in the mainframe world don’t install it without having run it through its paces. They typically don’t do it in the PC world either unless they are a small “mom and pop” business that had their local computer tech expert convince them to do so. One reason why so many Windows users are still running XP.
z/OS has been running on top of a hypervisor since the early nineties (PRSM/LPAR aka VM/HPO). Before then MVS could run as a guest machine on IBM’s VM opsys. The majority of the work to get z/OS off the proprietary hardware would be to develop a hypervisor for the underlying hardware. This has already been done by IBM. Currently working on a project to bring up zPDT as a “guest” on the cloud. Hardware is Intel. Software is IBM.
This is what I said. The migration path for those that don’t understand the value of the mainframe hardware and are doomed to regret it, is to emulate the hardware. The OS, program products, and applications cannot be ported or rewritten
We had similar rumors start low and get louder until finally my division was sold to Global Foundries. I think we are better off for it. We were an afterthought to IBM, now as long as we stay profitable, we are valued.
I hope the mainframe workers get to go to Hitachi, assuming the company plans to keep manufacturing in the USA. (assuming all this plays out of course)
“…Power chip in a mainframe”
Power chips do not run mainframes.
Do your homework next time.
So funny to read this! Have you ever seen someone that sold a monopol business? IBM live from the great margin the mainframe business delivers.
And most funny: porting the OS to Intel. To the one who wrote this: you have posted to the world that you know nothing of the IT world. Just think of this: why is a mainframe still alive and customers willing to pay millions? Why are still 80% of business data on this machines?
Bus again: thank you for that article – it saved my day and still puts a smile on my face!
All the talk about maintaining and enhancing the “mainframe architecture” is just that. TALK. The reality is that the mainframe community (developers, customers, investors, IBM itself) is dependent on an aging platform architecture that has lost many of its technological foundations. For every “mainframe developer” that exists today, I can find dozens of developers who work on more contemporary platforms, not to mention the major cloud platforms (e.g. AWS, Google Cloud, Azure). The skills are disappearing, and they will continue to disappear the way IBM has handled developer training.
Another thing that isn’t discussed much is the missing program product source code. IBM mainframes are well known for their backward compatibility…WHY? Of course it’s great to say that the latest Z system will run older mainframe code, but the fact is that a lot of the source code for those apps no longer exists. The source code is gone, and the developers who wrote that code are either retired or dead from old age.
The bottom line is that “the mainframe” is a legacy platform. It will sell for as long as it continues to run old apps and the required supporting software (Z/OS, DB2, CICS, etc.) If the mainframe business is truly for sale, then Hitachi or whoever will be taking on a stable maintenance business. It will be a money maker to be sure, but future application development will take place on other platforms. IBM could actually be helping themselves (albeit as a smaller company) if they manage this right.
“The reality is that the mainframe community (developers, customers, investors, IBM itself) is dependent on an aging platform architecture that has lost many of its technological foundations.”
That’s completely false. There’s more tech and more useful tech in the newest mainframes than any commercial computer available in the history of technology. Your comments smack of penis-envy.
“Another thing that isn’t discussed much is the missing program product source code. IBM mainframes are well known for their backward compatibility…WHY? Of course it’s great to say that the latest Z system will run older mainframe code, but the fact is that a lot of the source code for those apps no longer exists. The source code is gone, and the developers who wrote that code are either retired or dead from old age.”
Well you missed it again. It’s not that source code is upward compatible but even the object code. We don’t need the source. The executables from applications continue to run forever.
IBM guy: “Penis envy”? Hardly. I was quite conversant with many platforms during my time with IBM, and look upon mainframes with a sense of admiration. The problem is not that the mainframe doesn’t have “cool stuff” in a technological sense — the problem is that many people no longer care. The platform is aging, developers are retiring or dying, and the reason the object code for many applications is upward compatible is because that is all that’s left — there’s no longer any source code to develop, hence no evolution. Unless there are huge changes in the community, so to speak, IBM-compatible mainframes will run bigger and faster, with an ever-decreasing number of applications. And although it may take time, those applications will migrate off the mainframe onto something else. To say “We don’t need the source…The executables from applications continue to run forever” is short-sighted thinking — the kind of thinking, quite frankly, that drives CIOs away from any platform. Don’t believe me? Talk to any CIO and tell them “the source code no longer exists, but we can just keep on paying IBM”. If they can do it (or pay someone else to do it), they will quickly port that application to something else, and that something else probably won’t be a mainframe.
Former, I think you nailed it.
The people in both IBM and customers making the strategic decisions are not engineers who can appreciate all the tech in a modern mainframe. The way I look at it, the costs of entry are lower with modern distributed tech, and it can be built on a skill base much closer to what runs the rest of your IT shop.
It’s not so much a question of what mainframes can do, as what *only* mainframes can do, and whether it is worth the cost of the technology.
It’s all about business outcomes. The technical underpinnings are a means to an end.
While I agree that technology keeps improving on the mainframe, I disagree with most of your other comments. While maintaining backward compatibility has always been an objective, there are times when it is not achievable. It could be something as simple as asking the customer to recompile to something more such as recoding an API. I understand that it may less of a problem today vs 10, 20, 30 years ago, but it can still occur. It even happens in the PC world with Windows – what ran on Windows 3.1 may not run on 95/98 or XP or Windows 7, 10, right on up the line. There are already complaints that certain applications that run on Windows 7 (or 8) won’t run on Windows 10. Plus whose to say that the source for some of that stuff is still around – especially if some small business paid some college kid to write a BASIC program to perform a specific task.
As a 37 year retired IBMer this is painful to consider, but the alternative might be what happened to Kodak, simply disappear trying to do the same old thing.
Kodak is also a software vendor. One of the clients I work at is in the middle of winding down what has been claimed a disastrous implementation of their software for images. It apparently hasn’t been a good relationship and the blame from the customer goes to the vendor. As I only handled performance aspects of it as part of a recent issue, I can say there are two sides to the story and I don’t get to hear much from the vendor’s side.
I find this entire article very unbelievable. Bob clearly does not know how difficult it is to run modern Z workloads on any other architecture. While nothing IBM does will surprise me, including selling System Z to Hitachi, porting or emulating it on another architecture and expecting it to handle the same massive transactional volume with the same level of reliability and security is not going to happen.
Bob, just what is your source for this article? I hope it was more than a couple of emails with no substantiation.
I can’t see this gelling with Bob’s argument in Seeking Alpha….that essentially IBM is all about creating an impression for Wall Street, and using the highly profitable “traditional” revenue (which he says is 95% profit) to make the new stuff look more profitable than it is, in an environment where (he argued a month or so back) the CAMSS transformation has already failed.
If you accept the notion that the current executives are financial engineers, then why would they throw away their most powerful financial engineering tool?
In this case, closing IBM’s mainframe development department in Moscow (Russia) seems reasonable…
And huge process of system migration from mainframes in Boeing also confirm this news
May I make a comment which seems to bear no realtionship to all the previous discussion but in my mind just adds weight to the premise that IBM is dead in the water. In my opinion mainframes are IBMs lastfoot hold in the serious enterprise business ie banks etc. If they sell this they could make some short term money but will lose what little relationship they have remaining with big corporate clients. I dont think wether a mainframe is faster, slower or what ever makes any significant differance when it comes to commitment and relationship with clients, they want to feel they have somebody at their back, when that somebody drops them for what ever reason they will have no reason to meet and discuss anything and that means no future business. Assuming they do sell the mainframe business what is left that has a real physical IBM badge on it and any significant history and special knowledge behind it? Zippo and thats where the revenue is headed in my humble, non share market opinion.
I always find it amusing that Intel’s proprietary chips are never referred to as proprietary.
IBM is able to sell only mainframe’s hardware business, but software one never.
I think HITACHI never be able to move mainframe’s software to Intel platform.
It’s impossible, because it’s never possible.
I read that IBM mainframe business is a $5 billion dollar business but no growth, Amazon AWS is $10 billion and growing at 50% so within a few years it should be a $100 billion dollar business which will dwarf the IBM mainframe business. That is what IBM hates. They missed the Cloud growth but its too late now it seems Amazon AWS and Azure are getting the most market share.
(Cough, cough.)
.
“Why Amazon Believes AWS Business Could Be Bumpy”
.
http://marketrealist.com/2016/06/amazon-believes-aws-business-bumpy-going-forward/
.
Constantly cutting prices may be good for acquiring new business but it’s bad for the bottom line. Having to build out expensive new data centers may be a requirement for servicing that newly acquired business but it’s also bad for the bottom line. A lot of customers who have moved stuff to the cloud aren’t necessarily so happy with the results, and don’t seem to be saving any money. Plus there are some potentially serious legal ramifications here which they just now seem to be grasping. And so on.
.
This particular rabbit may be be running very fast now but it may drop from exhaustion soon enough. Don’t say you weren’t warned.
.
Wow, a whole lot of contrarian thinking going on here 🙂 You’ll win every arguement if you use logic like “A multi-core Intel processor is as powerful as a Power chip in a mainframe, and vastly more powerful than the mainframes IBM made in the 1990’s”.
Truth in advertising…I am a current IBMer and am responsible for IBM’s Cloud Managed Service offering for Mainframe workloads. Yes, that would be a Mainframe in a Cloud.
[…] Click to have full access to this article […]
I was surprised when I read the article. Like others here in the comments column, I’d question why IBM would consider selling off the core profit-making part of their business, but I think some of the commenters have hit the nail on the head. IBM has realised that while the mainframe still has a foothold in large organisations, many have been successfully migrating to lower-cost Intel platforms for the past 20 years, and this will only accelerate. IBM’s thinking may be to get as much as they can sell it for while it still appears to be an attractive purchase. The longer they wait, the less they’ll get for it. They probably don’t want to continue to invest in it; either that or it is becoming exceedingly difficult to enhance it due to the exodus of resources with mainframe software skills and a lack of newcomers learning mainframe software development.
The issue of performance is not really part of the argument. I agree that the mainframe reigns supreme with throughput – the performance of Intel servers on batch jobs is glacial compared to a mainframe – but organisations are designing their applications to work around this. I worked in mainframe environments for 18 years and on SPARC and Intel for a further 20 years. I’ve seen large software (porting) migration projects fail because the Intel/Unix platform couldn’t come close to the mainframe’s overnight batch processing (and watched the legal consequences play out), but I’ve also seen other projects succeed when the system is redesigned from the ground up. In the end, the cost savings in mainframe software licences and maintenance outweighed the redevelopment. Then again, I still see organisations putting up with some poorly performing and unreliable applications deployed on Intel/Windows Server. I think Bob’s prediction is probably right. I also think IBM’s chances of surviving as a Cloud company are slim.
The mainframe hardware and software business is about $10B in revenue.
Let’s say 70% of that is profit – it’s higher for some of the software (90+%) but much lower for hardware and z/OS itself
Typical buyouts are 5-7 times revenue.
So Hitachi pays $50B (IBM never gets good value for spin-offs )
Assuming IBM is making $7B/year in profit, that’s 7 years of profit.
Ginni is probably weighing where she thinks the mainframe market will be in 7 years, and what cloud/analytic/mobile stuff IBM could acquire for $50B *today* before the alleged window closes.
Assuming it would not all be wasted on share buy-backs.
Also, the RA’s have been savage in the mainframe LOB and the talent attrition is diminishing the value of the business every year. Best to sell now while it’s still worth something.
Ginni will be gone in 1-2 years, retirement – she’s 58 and usual CEO retirement age is 60.
She may want to leave as the CEO who truly ‘transformed’ IBM and take this huge gamble. Clearly she begrudges every dollar spent in this ’empty revenue’ LOB, so I could see this happening.
Hitachi I think would be fine, as would mainframe customers. They could cut the useless IBM management layer, put some $ into developing and sustaining the infastructure, and likely have a few more decades of a very high profit margin business.
The best justification I’ve read so far and it sounds like a reasonable plan. However, when it is very likely to be spent on buy-backs and suspiciously increased rewards for those at the top etc. …. Then it becomes a bad plan sadly. Fingers crossed there is a secret plan we haven’t heard about yet that will turn things around and this divestiture will fund it.
Any thoughts to the future of IBMs remaining storage portfolio ( v7000, v50x0, etc) ?
If mainframes go, then I wouldn’t expect any other hardware to stay either. This includes all remaining servers (i and p, or whatever they are called now), storage systems and anything else. IBM’s path for some time now has been to proceed on what will bring in the most money for the least amount of resources expended. This effectively means software, and to a lesser extent services. The staffing and other resources needed for a software and services operation is different from the resources needed for a long-term manufacturing operation. IBM’s management has known this for some time, but it seems with this latest rumor that they may be finally ready to admit it.
Good luck IBM.
We retirees had a great career and lots of fun.
SO IS IT A BUY, SELL Or HOLD ???
Only time will tell !
İşadamı Sadık Özgür tarafından kurulan ve Türkiye’nin önde gelen markalarından olan, sektörünün öncüsü Kale Kilit, hizmet ağına yeni halkalar eklemeye devam ediyor. Dev markanın bünyesinde bulunan gayrimenkul stokları yeniden değerlendirmek amacıyla 2011 yılında Sadık Özgür’ün kızları Sema Gürün, Sefa Çizer ve Sedakat Özgür tarafından kurulan 3S Kale Gayrimenkul Geliştirme ve Yatırım A.Ş. ve 3S Kale Yapı ve Ticaret A.Ş. şirketlerinin yeni yol haritası kamuoyuna açıklandı. 2014 yılında 3S Kale Enerji şirketi ile jeotermal enerji alanında yatırımlar da yapan grubun 3S Kale Holding A.Ş. çatısı altında toplanmasının ardından her iki alanda da büyüme hedefleri olduğunu belirten Yönetim Kurulu Başkanı Sema Gürün ile kız kardeşleri ve Holdingin Genel Müdürü Alp Gürün (oğlu), aralarında Yayın Danışmanımız Mehmet Canıtatlı’nın da yer aldığı bir grup gazeteciye şu açıklamaları yaptılar:
Avcılar Konut Projeleri ile sizlerleyiz.
https://www.emlakdream.com/haber/3s-kaleden-topkapi-ve-avcilara-konut-projesi/77689
Mainframes are a dying business, although still a very profitable one. But it’s a business without a future. Sell it while you can still get a premium. IBM is trying to transform itself into a cloud/analytics business It is wisely ridding itself of legacy distractions.
I’ve seen this movie before..?! I find it amazing how totally different companies are so adept at destroying their success..!
At one time I worked for Douglas Aircraft in Long Beach, Ca. They were a ragging success. But then ‘junior’ took over the company and ran it into the ground. Hence, McDonnell bought them out and became McDonnell Douglas. Again, a ragging success. Until, again, another offspring took over. In just a few years one of the last ‘big’ aircraft companies went out of business and sold to — of course — Boeing aircraft.
So what’s this got to do with IBM and computers you might ask?? A LOT! I watched both companies do the same thing IBM has been for the last several years. Planning on the short term and not relying on the talent they have within the company killing innovation and ruining corporate profits. Instead of listening to their subordinates, the elite at the top primarily follow the ‘in crowd’ (competitors) and do what they do. But we all know that reinventing the hoola loop ‘second’ doesn’t work very well.
Having spent over 20 years with IBM, I am saddened to say I am no longer proud to have worked there…
The problem is that this is now a SOFTWARE and PROCESS problem and not a hardware one.
The only reason the Z architecture still exists that customers still run software on it. Reliabily, Dependibly, Scalably, with Stability) Period.
Software like bank treasury, check processing, ERPs, airline reservations, package delivery.
Executives at most large companies STILL BELIEVE that the mainframe is the only platform with enough scale to do these jobs.
From a pure hardware perspective, that is no longer true. Google and others have proved that Linux clusters can be built at large scale with wave multiplexed fiber networks (bundles of fibers with very fast switching hardware) at 200gig/sec. Even 10gig/sec Ethernet on Cat 7 twisted pair is as fast as most processors. With enough static memory, these companies COULD solve these problems on this new architecture instead of Z. A Linux cluster can “easily” handle an airline reservation load of 5000 messages a second, or even a credit card messaging network.
Why don’t they? Well, first of all, the “abilities” (stability, reliability, dependability, scalability). They have been on the mainframes for decade and the latest generation have had precious few HARDWARE failures and can support five nines and have been proven to do so. IBM spent a fortune to PROVE that to these companies in testing labs. IBM earned that right through 3 decades of “abilities”. Much easier to manage your company with one vendor (IBM), than to argue about how a Dell server hiccuped with Red Hat Linux and a Cisco switch. Why are they so expensive? I think they are so expensive because of exactly that. IBM has spent a lot to PROVE that they are reliable enough to run an enterprise, while Google has not built and funded such tests. Mainframes cost more because more testing is done to prove their “abilities”. Now, if IBM takes shortcuts on that development and testing (which some here would argue that they are), they will produce an “unreliable” generation of Z and then the party will be over. Customers will no longer be able to trust IBM’s “abilities” and then the party will be over.
But rest assured. The customers are EXPECTING the next vendor to provide the “abilities” (stability, reliability, dependability, scalability). And you want them to. You don’t want your bank losing your bank balance, or the ability to handle payments from VISA, MC, AMEX, etc. You don’t want your airline reservation systems down for hours at a time when you need to board a plane, or book a flight.
The customers are slow to change, because moving to a new architecture is both dangerous for them, and expensive. And, in most executive’s minds, the benefits are probably not proven yet.
Yet, here come these pesky Internet startups. Bitcoin, Venmo, Crowd Funding. Will they provide the “abilities” on the new platform? Will people require the same level of integrity as a bank? Will Venmo be subject to “Know your customer” and money laundering statutes?
The next revolutions aren’t hardware. The big companies are looking at replacing mainframes. Some have. IBM possibly shopping the Z part of the company is a tacit admission that the time to retire these systems is near. I worked on “SABRE Labs” projects in the 1990’s which started to look at replacing the Z airline reservations systems. In the minds of the executives, the technology either wasn’t there, or they didn’t trust a vendor as much as they trusted IBM on the “abilities”. It was (and still is) a “research project” and when the money goes tight, the management did the “go with what you know” thinking and shuttered multiple generations of mainframe replacement projects as a cost saving measure. I’ve seen it at MANY other companies as well.
And this is where it is getting dangerous for IBM. With the BIG foray into Cloud, what do they do with all their “on premise” and “Z” business? They want to convince big companies that the cloud is as good as “on premise”. Guess what? I think precious few banks, retailers, airlines, etc. are prepared to give that business up to “Cloud”, and if IBM makes them, they will seek another vendor. IBM might run it for them at their premise (and that is what GBS/GTS does a lot of), but losing sight of the processing to an unknown place (like AWS does), even for a short time, is simply NOT going to happen.
The “abilities” were what has historically given IBM the fat margin. It translated into customer confidence in IBM’s people (now being laid off), their support (now being outsourced), their products (now being shopped to sell to other companies). Customers could trust that IBM would “make it work”, “make it right”, and that their “field service” guy would take care of them when any hiccup happened. Old timers complain about the symptoms, but the real disease is losing the trust of your customers in the “abilities”.
Companies are more concerned with these “abilities” than they are the specific hardware platform. Find the company who will handle this and test and keep the same people in stable jobs, and you have the next high margin, stable company who will run the infrastructure of America. And they should be well paid for it.
No way. Absolutely no way. IBM makes the bulk of its profits on mainframe software. It will be the last thing they sell off in the company. It is like a rumor Oracle will sell off Oracle DB…. Maybe they outsource the hardware manufacturing or something of that nature, but the software stack. Not ever.
So, last year you told us that IBM would fire 25% of its staff.
This year you are telling us that IBM wants to sell the one crown jewel … you know, that one single cash cow that continues to add billions and billions of revenue, and more importantly profit.
And I really don’t get what you mean by comparing a modern Intel processor with the capabilities of 1990s mainframe processors. Or was that sentence just a shortcut for “I have absolutely no clue what a year 2015 mainframe processor is doing and capable of”?
Long story short: it seems that you do not understand how important the mainframe dollars are for IBM. And just to be precise: of course, there are a lot of managers out there that focus only on Dollars and cost. But especially those guys do understand the value of a solid cash cow.
Bob didn’t start the rumor, he just says he wouldn’t be surprised if it were true.
Is Cringely actually Stewart Alsop in disguise? This has got to be one of the most ridiculous things posted on the Internet in years. Why would any corporation sell the line of business that is responsible for generating so much of its profits? The software, services, and everything tied to the mainframe is basically a recurring annuity stream for IBM. Wall Street would revolt and Investors would not allow the CEO to keep her job. It would be the last straw.
Does anyone think that the clients who for years have run their core business on the mainframe are simply going to say oh ok? Corporations have invested trillions of dollars in mainframe applications. They do not have the time or resources to port those over to another architecture. Especially en masse.
If anything, IBM would be wise to dump the non mainframe related businesses and go all in and focus on z Systems.
“Pundits” like Alsop and the author publish outrageous things like this to generate publicity for themselves. Did anyone else notice how the website showed a pop up add from Amazon for Cringely’s last book?
Re: “Why would any corporation sell the line of business that is responsible for generating so much of its profits.” The reason would be to maintain the eps. To do that they have to raise revenue or cut costs, to keep earnings high, or cut the number of shares. They have done the first two already, now they have to sell valuable assets to raise money for share buy backs. The revenue can’t increase if the customers are leaving for other alternatives, but they can always easily fire people or sell assets, thereby maintaining the eps ratio as the company winds down to earning $20 for one outstanding share.
Never underestimate the power of stupid people in large numbers. The executives don’t manage the corporation, the media demands flashier results, and they respond to keep their ‘good name” in the opinion organs, thus guaranteeing their own profit at the expense of all the “stakeholders”
And remember, customers and employees aren’t stakeholders – only a few major stockholders, correspondents and executives, are “stakeholders” everyone else be damned. They respond to the culture, they no longer create it.
IBM’s only remaining product is financial engineering – There was a time when a former CEO (I think it was John Opel said that IBM “wanted to be the GM of the information industry” – well, they made it – well on their way on the same performance curve for all the same reasons. By saving a few cents per car, GM alienated its customer base to the point where repeat business was no longer the norm. What did Bill Durant know that Roger forgot? Likewise, what did “Junior” know that Three Fingered Louie” and the successor vandals were never able to comprehend?
Sad, but predictable.
I think you are wrong there. Of course managers understand the pressure from public; but as you said: the huge shareholdes … those people understand numbers.
Thing is: if you take away the mainframe revenue and profit … you would shrink IBM by at least 30%; and regarding profits … probably by more than half. The hardware is just a small part of that, but tons and tons of stuff on top of that makes this platform so profitable to IBM.
And you know: Google … how often do they talk about “search” in public? Probably … not too often. They like to talk about cars, and glasses, or other kewl gadgets. But surprise: in the end of the day, Google is about searching. That is where the money comes from that drives everything else. So, do you think Google would give up searching to turn into a car company?!
What will bring IBM down is it’s own management. From the lowest employee up to Ginni Rometty there are 10-15 levels of managers. What do these managers do? Come up with ideas like this one?
Certainly they are not producing any code, nor chips, not any other item they could make profitable.
They just shuffle resources around, hire and layoff worker bees, etc.
I mourn for the IBM I joined in 1966. Not so much the black shoes and socks or the white shirt and rep tie. I weep for the lack of any compassion or mercy today. The head honchos in Armonk are driven by one goal – more and more money for the shareholders, neglecting the people who drive the engine that provides
the money. And indeed, totally neglecting retirees. We were assured that as DB pensioners, we’d be provided
with periodic ad-hoc increases in our pension. I retired in 1992 and have been awarded ONE pittance of an increase in the 24 years since then. Is it any wonder we regret the passing of old IBM.
At the federal level, the Prime Minister of Australia is blaming IBM for our Census failure on Tuesday night, 9th August 2016.
https://www.news.com.au/technology/online/security/its-not-the-first-time-the-government-has-been-upset-with-ibm/news-story/a9b08f22cd7582af595a201285c67840
At the state level, the ex-Premier of the state of Queensland chipped in to remind everyone that IBM also screwed up the technology providing the backbone to our Health system, just a few years ago.
https://www.itnews.com.au/news/queenslands-ibm-ban-lives-on-420969
Big Blue making people feel blue and becoming battered and bruised (blue) itself, as a result.
Another hilarous comment from Cringely. The fact that he says the mainfraimes have Power processors in them tell a lot about his knowledge about the Z/mainframe systems. hahahaha
[…] IBM’s server hardware business followed to Lenovo in 2014. There are rumours that IBM will even exit the mainframe business, potentially selling that to Japanese firm […]
I have been an IBM employee since the mid 60’s when IBM was effectively the dominating computer company. There were a number of IBM principles the company followed, such as “respect for the individual”, and “IBM=Service” among them. The current company has lost everything that made it great those years and now the only respect left is the respect for the Wall Street investors and the salaries of the top management.
I was in Las Vegas talking to a young man selling timeshares and found he had never heard about a company named IBM…. what that tells me is that the company is in freefall and ready to disappear into obscurity. Top management is only concerned about stock price and top management salaries – the average employee is now a liability – not an asset. The only positive point I can make about the mainframes though is the architected security in both hardware and software which is unparalleled.
I do not understand top management thinking – there should be no reason that stops the company from competing intelligently and effectively in areas like the cloud. After all, I remember a “white paper” decades ago that proposed something very similar… it just did not get sold as the sales reps did not get any real money selling it… In short: I am very worried… it is not a nice story.
IBM without the MF hardware will be much like what CDC became. Remember when CDC was significant? Not worth the effort to think back that far, is it.
Hopefully Hitachi will be a bit smarter with z architecture than Oracle has been with SPARC.
Platform Solutions Incorporated built a working z/OS interpreter using Itanium chips/assembler, running with RHEL as the I/O supervisor in 2005. Ran on Superdomes and smaller Itanium based workstations, about a 50 MIP uniprocessor. IBM subsequently bought the company. No software porting, just the performance issues associated with interpreters. It was a great product, but it was designed to run on commodity hardware, not bulletproof IBM/Fujitsu/Hitachi hardware.
In early 80’s, I was in Raleigh with customer discussing SNA, VTAM, and NCP. Customer asked why MVS just support TCP/IP natively instead of SNA where everything must be pre-defined and pre-configured. The product manager proudly said “Not in my life time”. Well, less then even 10 years, no one talk about SNA anymore. I understand that MVS is cash cow but there are so many arrogant IBM STSM, DE, and management never see what’s in front of them.