blog Due to the incredible rise of the x86 chipset over the past several decades, it’s easy to forget that at the beating heart of many organisations, those ancient dreadnoughts which the IT industry knows as the common or garden mainframes are still found, continuing to promulgate their ever-lasting mission of stability, cost efficiency and power. And some organisations, such as the Northern Territory Government, are even buying brand new models. From an IBM media release on the subject:
“Sydney, Australia – 12 Jun 2012: IBM today announced that the Northern Territory Government has selected the IBM zEnterprise 114 mainframe server to host a range of governmental business applications.
“We have a longstanding, successful relationship with IBM, which is recognised as a leader in technology innovation,” said Bob Creek, Director of Data Centre Services, Northern Territory Government. “Due to the multi-access, shared nature of the government’s systems, the IBM z114 upgrade will ensure our mission critical government applications and systems run as efficiently as possible. The mainframe is fundamental to our business and we believe it will continue to deliver efficiency benefits to government for a long time to come.”
The computing horsepower of the IBM z114 ensures the Northern Territory Government can run its services seamlessly and securely. The IBM z114 is an especially attractive option for organisations experiencing rapid growth in new services. These organisations are faced with ever-increasing torrents of data, and want smarter computing systems that help them operate efficiently, optimise decisions in real time and reduce risk.”
As a technology journalist, I’ve been hearing about the imminent death of the mainframe for more than a decade now. First Sun’s SPARC processors were supposed to kill them off, then Intel’s Itanium architecture was supposed to kill them off, then Intel and AMD’s x86 servers (especially the blade servers) were supposed to kill them off. And now, I guess it’s the Oracle-branded Exalogic and Exadata machines which are the big kids in town, although of course their processing purpose is somewhat tangential to the mainframe mission.
But throughout that time, the mainframe has survived and is continuing to thrive. I recall discussing with CSC Australia’s then-managing director Mike Shove about five years ago how it was hard for the IT services giant to get as many mainframe developers as it needed, despite the high level of continuing demand for such skills. Mainframes weren’t sexy then, and they’re not sexy now — but they get the job done. And I continue to hear rumours about large chunks of IBM Australia’s revenue being based on mainframe work.
An interesting point regarding the NT Government’s mainframe deployment is that I believe IBM may have pipped Fujitsu (another major mainframe player) to the post here. Fujitsu, after all, has long enjoyed a strong relationship with the territory with respect to its mainframes, inking a major contract for mainframe application services back in mid-2007, for example. I wonder if Big Blue is cutting Fujitsu’s grass somewhat in this new deployment.
Image credit: IBM
Well, that should warm up their data centre…
I don’t think they have a problem keeping things warm in the NT :)
Hardly.
Big Iron mainframes still rule the roost for ultimate performance, horizontal scalability and stability. IBM Power and Z systems count unplanned outages in milliseconds-seconds per year as opposed to x86 architecture which counts in seconds-minutes (i.e. 99.999% uptime equates to 5.26 minutes of outages PA…not a big deal unless it’s an air-traffic control system, heart-lung machine, multi-billion dollar currency processor, etc….)
Sure, mainframes still have a place; however, it’s not in Darwin. A little local perspective:
1) We don’t need the scalability. The NT has a population of only about a quarter of a million people, and many of the applications which were previously running on the mainframe have in recent years been successfully migrated to mid-range systems – so the remaining workload is minimal. The few applications that are left could easily be handled by a handful of additional mid-range servers.
2) Nor is high availability relevant in this case. Even putting aside the fact that the NTG mainframe doesn’t perform the kinds of roles you mentioned (people run their heart-lung machines from mainframes???), the constant availability of the mainframe itself is moot when the data centre it lives in suffers frequent network outages lasting minutes to HOURS (and near constant hiccups lasting several seconds).
The only valid reason for this purchase is to buy a little more time.
Don’t break what’s not broken. Because fixing an I.T. service while its in production is like fixing a plane while its flying…
I watched a company who shall not be named *cough* *cough* medibank burn 1 million bucks a day on trying to get rid of mainframes, no idea how it ended up, probably badly.
The z114 is a highly configurable system (much more than your typical PC). For example, it COULD be running a bunch of Microsoft Windows servers on x86 chips. Or it COULD be running a bunch of Linux servers on POWER7 chips (descended from the old Apple PPC), so in a way the mainframe architecture did die, but in another way it merely evolved and adapted to meet modern expectations.
If you were going to go the other way, you would probably buy a bunch of DELL or HP machines, stack them up in a big iron rack, put in plenty of fans and some sort of management system for remote access, and you get something kind of similar to a mainframe anyhow (but with less exotic CPU and IO options).
That’s the whole idea of computers — they are general purpose devices.
Given the amount spent on non-mainframe elements of the NT Gov IT infrastructure (i.e. $150 million desktop deals with Fujitsu), the spend on a baby mainframe is miniscule by comparison, likely much less than 1% of their desktop project. It provides an exceptional price/performance point when compared with scale out x86 based kit. Unless you have direct experience comparing the costs of running both, it’s just not possible to appreciate what this system does for them.
@Zwan – that company still has their mainframe – and they are not alone in spending hundreds of millions to unsuccesfully move to an alternative. A centralised payments processing organisation that interfaces with every bank in Australia is in the process of upgrading an more than 10 year old out of support mainframe system because they thought they could replace it for chump change – over $100 million later they are realising their mistake. But by now the decision makers have moved company into nice new shiny roles where they can attempt to force their biased anti-mainframe point of view on yet more unsuspecting Boardrooms – all to the ultimate cost of either the shareholder or the taxpayer…
Comments are closed.