If you believe the stories, IBM's mainframe has in the last 10 years been knocked down and gotten back up more times than most of the characters in all the Terminator, Die Hard and Rocky movies combined. And while there are some out there who'd like to see its demise, a true threat to the Big Iron has never really amounted to much. Even today, the proponents of commodity boxes offering less expensive x86/x64 or RISC technologies say the mainframe is doomed. But the facts say otherwise. In fact it seems like a pretty good time to be pushing mainframe technology.For example here are a few positive facts:
* IBM recently said the mainframe has achieved three consecutive quarters of growth, marked by new customers choosing the platform for the first time and existing customers adding new workloads, such as Linux and Java applications.
* Mainframe MIPS growth is averaging around 20% per year and large mainframe-centric enterprises have been consistently averaging 35%-plus MIPS growth, says the Ovum consultancy.
* The realities of large investments in core application deployments on the platform and its long-heralded reliability, availability, serviceability, security, backwards software compatibility, efficient environmentals and virtualization capability, have rendered negative mainframe prognostication moot.
* IBM said will spend $100 million over the next five years to make the mainframe easier to use.
* IBM took the path of providing a control structure and very attractive pricing for using System z mainframe processors for new workloads. For example, when purchasing, or upgrading a System z, a customer can request one or more processors for exclusive use by workloads running under Linux. These Linux-only processors are priced approximately 90% lower than a processor utilized for a traditional mainframe workload. The result is that using System z virtualization, a customer can configure a System z at a very attractive price with the capacity to run workloads from literally hundreds of discrete i86 servers. IBM also offers similar special pricing for processors in the System z for data base workloads and for new types of workloads such as gaming, says Saugatuck Technology researchers.
* Saugatuck says IBM and Unisys mainframe platforms have succeeded in a key area where non-mainframes have failed. Both companies enable non-mainframe workloads to benefit from many of the capabilities which had been exclusively experienced by traditional mainframe workloads. Specifically, both approaches provide levels of security, integrity, availability, etc. which each vendor has engineered for the past four decades into their respective platforms. * IBM said its hardware mainframe revenue grew 12% in the first quarter of this year compared to the previous quarter, and 25% year-over-year. The strong growth is partly due to increased mainframe deployments overseas. IBM's mainframe revenue grew in Latin America and Russia, and it doubled in India and China over the past year. A company can run its entire business from a single mainframe, IBM says.
* IBM has confirmed that approximately 25% of its System z capacity currently being shipped to customers is designated for Linux workloads. Further, IBM says the System z has been gaining high-end servers and that System z capacity shipped in 4Q06 was greater than the total capacity of the then current installed IBM mainframe worldwide inventory.
* IBM touts a successful education program in its ZNextGen offering. zNextGen, a community that IBM and it's large user group Share worked together to launch late last year for new IT professionals focused on mainframe computing. IBM says 42,000 students at 328 colleges and universities have been educated about mainframes. Additionally, IBM created an online meeting place, dubbed Destination z, where customers, system integrators, software vendors, and academic institutions can connect with experts, access development tools, and the latest mainframe software. BMC Software and CA also have spearheaded programs to strengthen mainframe-related training.
* IBM developed the Cell Broadband Engine chip in conjunction with Sony and Toshiba for use in industries ranging from aerospace and defense to gaming, including Sony's PlayStation 3 video-game console. IBM intends to create a hybrid machine through the integration of the Cell processor with its System z9 Business Class (z9BC) mainframe server. The z9BC is the vendor's entry-point product for businesses that need less capacity than the System z9 Enterprise Class mainframe.
Layer 8 is Network World's daily home for the not-just-networking news.
Contact Layer 8
Layer 8 archive
|
|
"...hundreds of discreet i86 servers..."
I guess they're modest, and like to keep to themselves. A security feature, perhaps?
Oh...! You meant "discrete".
Never mind.
No significant increase in new workload
There has been no significant increase in new workload on z/OS, only an increase in existing workload. The increase in MIPS is because current customers upgrade their current hardware or increase in their current workload. They do not want to change the software. z/OS is both arcane and arcaic and no prospecting customers will want to invest their IT dollars in it. The mainframe hardware is very reliable and zLinux is the new thing, not z/OS. z/OS is hard to maintain and very time consuming to work with. there are very few skilled workers with most of them in their 60's, and only few young workers want to work on it. IBM could win over more customers by making a free visual interface that runs on Windows. Uptimes are measured in hardware reliability but why not in software? How about adding downtime caused by user errors. You veterans can testify to that. Moreover, there are very few mainframe job openings.
My friend, you are plain
My friend, you are plain wrong.
I work as a system programmer in a large bank. In my team, besides our manager, there is only 1 person over 60, and 2 people (including myself) under 30. Moreover, there is more job demand than people to do the work, even in a small country like I am from. Have you actually checked lately for job openning in the US or UK - you'd be surprised.
And finnaly, I dont have numbers, but fact is almost every MF site I know is currently in some phase of implementing Java, WebSphere, SOA, etc. That is what you call "new workload". The classic workload still has much value to the organizations who keep it, otherwise they wouldnt have. Its no coincidence that every major organization in the world runs z/OS. It's what manages your money, its what builds your car, it's what brings electricity to your house.... So maybe it doesnt have messenger running on it, but where would we be without z/OS...
cost prohibitive
The reason there are so many mainframes still being used is the fact there would be little or no return on the investment necessary to move off the mainframe. I have recently moved from an Oracle environment to a project that uses mainframe db2. I have NEVER received as many support calls as I do now.
z/OS is both arcane and arcaic
I can remember using OS/390 2.5 and subsequent releases during the mid 90's when LPARs would run more than a year without being IPL'd and that included applying updates to the OS on a regular basis. I know that MS and Linux OS’s can not sustain an operational period near a year or two without a reboot as a result of maintenance. Seems to me that the IBM OS (that’s software not hardware) is quite stable and maintainable without interruption.
free visual interface that runs on Windows. They call that IE. Which real Unix or Linux systems programmer would use a GUI to do their OS work? Oh, did you mean the other installed software? Well the GUI interfaces are available from various sources just like for Windows and Linux.
How about adding downtime caused by user errors. You veterans can testify to that Sorry, as a veteran I can’t really agree with you as I can’t recall a time when a USER brought down the OS short of the USER actually being a systems programmer.
Moreover, there are very few mainframe job openings. As far as jobs go I suggest you try keying in an IBM product name at Monster.com for the USA and see how many hits you get. Key in websphere You can’t read all the listings in a week.
I thank you for writing your opinion as it provided an opportunity for me to voice mine. I too would not have written mine unless I think I am correct. Thanks.
Pete
Re:z/OS is both arcane and arcaic
Sorry, as a veteran I can’t really agree with you as I can’t recall a time when a USER brought down the OS short of the USER actually being a systems programmer.
Excepting the time the SysProg left the default JOBCLASS at A and I forgot to code a JOBCLASS statement. C program, running a loop of mallocs
over the employee list... Oops.... but I did get to meet the DP VP that day..
8=o
Confusing the Issue
The idea that new apps are not being developed for the mainframe is deceptive. The new BUSINESS apps being written on little-iron are not new at all! The new developments on little-iron are graphic versions of old interfaces. Rather than replace any real mainframe function, they have simply increased demand for mainframe back-end systems.
This is difficult to accept in a culture which has enjoyed being the god of its own little desktop. REAL systems are a collaborative effort based on economies of scale, a concept well understood by big business. So pervasive is the new-age marketechture that many young workers have come to accept it as inevitable that the future belongs to those who can type with one finger. However, when the see what mainframers get paid, their impression changes very quickly.
Besides being vastly more economical in all but the smallest operations, mainframes are also orders of magnitude more reliable, going months on end with neither scheduled or unscheduled outages. Furthermore, the error recovery alone in z/OS is bigger than all of Windows, Unix and Linux put together.
When you see a dog dancing, you don't say, "my, what a good dancer!" You say, "it's a wonder the dog can dance at all." Rather than fight success, why not spend a few weeks in a real computing environment and see why the mainframe is bigger than ever, as the article indicates? When the latest fad PC OS is history, the mainframe will still be around.
Arcane and arcaic? Many people would think you meant UNIX.
I have pesonally heard hundreds of Windows support people say exactly the same thing about UNIX in all of its various flavors. After all, there are probably a thousand or so Windows users (at least) for every person who knows anything at all about UNIX. That doesn't make UNIX a bad OS, be it zLinux, Red Hat, Novell, Solaris, HP-UX, AIX or any of the other variants.
z/OS, like OS/390 before it, has Unix services which allow many UNIX programs to run, either C++ or Java, or even (gasp!) UNIX COBOL. WebSphere is a J2EE server, which is where most application programmers' connection to the OS stops anyway, so what's the point. And as someone else has noted, z/OS is still more reliable. One study placed the mean time between an OS crash in z/OS at 30 years, as defined by an event requiring an unplanned IPL. Show me anything else which even approaches that and I'll be interested.
It's called OpenVMS
I love this thread. It's great. At least some people are talking and voicing (some good) opinions. You want an operating system where uptime is measured in years? OpenVMS. It's been out a long while. As a system programmer/admin and consultant with experience from 1979, I've used just about everything. Some I'd rather not own up to. IBM 4300s running VM/370 CMS, VS1, VS2, MVS, yada. Tandem Guardian, VAXen with VMS from 4.0 (to the current 8.3 on Alphen). I've microcoded VAXen with user microcode, written IBM Assembler and PL/1 until I could read punch-card holes in my sleep. I've twiggled (it's my word) IMSAI 8080 front panel rocker switches to write a program and booted my (first) Apple ][+ in ways Apple would cringe over.
I have a client that uses an old VAX 3800 running VMS 4.6 (really old) to control their steam boilers of a BIG, HUGE earth-moving tractor builder. Last time I was there and worked on the system I had to reboot it (bad hardware). That system had run for 4+ years without a boot. 24/7 system. That was in uhmmmm 1997. I've never heard from them again about another issue with it.
Seems people forget about an operating system that is secure, stable and scaleable. Must be because Compaq (now HP) has their head so far up their.... Okay, well at least HP managed to port to Itanium2 and maintain OpenVMS on Alphaens. Hopefully they really will keep it alive. It's still used in a lot of uses across the planet. You want a stable OS with security and high uptimes? Pick it.
Mainframe is Green!
Another reason why the mainframe is gaining popularity is that it is a Green star. It is very energy efficient compared to running workloads on 100s or 1000s of discrete servers. Many studies show the mainframe to be an order of magnitude more energy efficient than running on the latest x86 servers. The virtualization technology of the mainframe is the most efficient in the industry, far more advanced than VMware. The virtualization z/VM provides z Linux normally allows 30 - 60 servers on a single processor providing great energy and software savings. Mainframes are the only platform designed for being run at 100% utilization and have great quallities of service. Centralized computing is making a comeback because it pleases both the CFO and organizational green initiatives by utilizing assets more, saving energy and reducing the power and cooling stresses that data centers are experiencing from x86 server proliferation. Unless your IT budget is growing, and most are not, then additional costs for power and cooling discrete servers impact new app development, manpower or the ability to keep the complex environment running that most shops have evolved into. Mainframes can bring order to chaos while doing it in a very green way.