As part of its Big Green initiative, IBM is replacing some 3,900 servers in its data centers with 30 mainframe computers – a move that the company says will result in a staggering 80 percent decline in energy usage. Ever more powerful servers were supposed to be the death of the mainframe but Big Iron is back – with a twist. Software will let each Linux-powered mainframe act as thousands of "virtual" servers, providing the same capabilities of individual physical servers while consuming significantly less electricity. IBM (IBM) will roll out the project at data centers in Australia, Japan and the United States. "The mainframe is the single most powerful instrument to drive better economics and energy conservation at the data center today," said James Stallings, general manager, IBM System z mainframe, in a statement. IBM’s greening of the server farm comes as tech companies like Advanced Micro Devices (AMD) and Sun Microsystems (SUNW) join Internet giants like Google (GOOG), Yahoo (YAHOO) and Microsoft (MSFT) to slash computer energy consumption and combat global warming through such programs as the Climate Savers Computing Initiative and the Green Grid.
Big Blue’s Big Green Iron
August 1, 2007 by Todd Woody
Green Wombat 
Gag me with this nonsense. If mainframes are so good why didn’t IBM use them to begin with and save themselves the trouble of decommissioning 3900 systems? IBM must have a lot of excess mainframe inventory they need to put to use.
Previous comment obviously made by someone who doesn’t have to buy, install and pay ongoing costs for dozens of unix boxes. By the way, most of those unix or NT box wish they could approach the uptime and economics of those “excess” mainframes.
Makes sense to me. I’ve toured many data centers. Mainframes don’t have anywhere near the footprint they used to while the Windows servers occupy racks and racks. Personnel is the same way – mainframe tech support groups are smaller but there is usually an army of windows engineers and administrators. Mainframes are good, it’s IBM’s once mighty marketing machine that lost this battle…
Likely because they can now run Linux under multiple logical images on a single mainframe, whereas before this was not possible.
What this article doesn’t say, but what is clearly part of the press release, is that the mainframes are running the Linux operating system. The readiness of this configuration wasn’t mature enough when the 3900 servers were built up over the years to start with mainframes in this way. But this says that’s not true anymore…it’s the combination of mainframe and Linux that makes this so compelling.
I’m amazed, amused, mystified, and perhaps a little stunned at the comments. I can’t believe IBM would mess with anything but their own products in their datacenters which would eliminate the MSFT crap. If they’re implying that they have 3900 AIX based or AS400 systems that they’re replacing with mainframes that’s a pretty pathetic statement about their new AIX/p6 based systems. As for Linux, is anyone dumb enough to run an Open OS designed for cheap hardware on the most closed & expensive operating environment on the planet? The last time I checked the overhead of running more than a few hundred Linux instances on a mainframe required the the Real Time Clock be slowed down to the point where the system time had significant drift. The mainframe simply couldn’t handle the timer interrupts Linux generates. Other than that which Linux ISV’s support their
apps on a mainframe. I’m sure all of the open source developers have mainframes in their garages to test on.
too funny.. what about the fuel costs of the 34,000 consultants it’ll take to get WebSphere up and running on these systems… you know things are bad when a ‘green’ angle is what’s required to sell mainframes…
Servers were the little engine that couldn’t. Big business knows that now and they’re going back to big blue. The big iron dinosaur’s will rule the earth again. :o)
Clearly IBM realized big iron can save energy and centralization has efficiencies and now is exploiting that fact where it makes sense. The mainframe a big server designed from day 1 for efficiently running many workloads at high utilization rates. Distributed servers have a design point that lends to inefficiency by attempting to scale out for performance. It’s simple, less hardware draws less power.
Skeptics, Why the hostility towards big iron? Add up the numbers before being critical. Green? Fewer efficient servers is much better than many underutilized servers. Common standards and middle ware have leveled the playing field. Large scale computing can easliy run many apps on fewer servers. Motorcycles get great mpg. Busses and trains get better mpg/passenger. Why should it be any different with servers?
Vendor lock in through consultants… IBM is running your software nothing can go wrong, can go wrong, can go wrong…..
Stewart, nice that you are mentioning problems that were fixed in 2003. As far as people dumb enough to run Linux on a mainframe? Lets see…American Express, Merryl Lynch, Blue Cross Blue Shield amongst many others. Yes, you are right. They are all idiots. Maybe you should open your eyes, check out what is happening in the real world, get updated information and then when you have educated yourself share your opinion. Until then, your opinion isn’t worth the bits and bytes needed to display it.
-Kevin
The full press release is http://www-03.ibm.com/press/us/en/pressrelease/21945.wss
With respect to the comment about what the move may say about AIX/p6 – actually it says nothing at all about AIX/p6.
Any company that wants to improve their overall energy efficiency is wise to look at consolidating in a way that makes sense for them. Some companies may choose to consolidate on AIX/POWER6 or HP or SUN or whatever, but with thousands of servers supporting hundreds of thousands of users, the move to mainframes makes sense from the economies of scale point of view. Could IBM have chosen to perhaps consolidate to fewer POWER6 based AIX servers? Sure, but would the energy savings in terms of raw power draw of those servers and cooling costs have matched the expected 80% savings from going with a mainframe solution? Probably not.
Some have commented with some very humorous remarks about linux and availability of applications for the mainframe distros. True, the average open-source coder probably does not have a z9 sitting in their garage, but likewise the VAST majority of ‘average’ open-source coders are not writing applications to handle day-to-day business operations for a Fortune 50 company. As for the question about ISVs providing software for Linux on z — do some research and you’ll probably find this link… http://www-03.ibm.com/systems/z/solutions/isv/linuxproduct.html
There’s a few 🙂
These trends – local vs. central, etc. – come and go, come and go, all for good reasons. For example, it makes sense to replace 3900 servers with a mainframe, but it doesn’t make sense to buy a mainframe instead of your first server, or you second server, etc.
The thing mainframes have going against them is IT management. CIO’s and their managers are justifying their salaries with convoluted pc server environments and the huge staffs required for support. Don’t get me wrong – we all want our email and pc’s have their place on the desktop. But for mission critical apps, mainframes mean smaller management empires. Most CEO’s don’t understand that they’re being sold a bill of goods by IT.
All the anti-Big Blue whining above seems to miss the point that they are realizing an energy-savings of 80%.
Folks we’re talking millions, maybe billions (if you’re Yahoo, Google, etc.) of dollars here in annual energy costs.
That is reason enough to merit investment in a mainframe.
if it means i get to run a lotus notes client – i’m in! what else can i run on that piece of big iron? sametime? WOOO HOOOO… this is living.