Monday, 26 March 2012

IBM takes the QI approach!

The BBC in the UK produces a wonderful programme called QI – IQ reversed, but officially standing for Quite Interesting. The point of the show, which is a humorous quiz show, is to not only illustrate how little we know, but also highlight the things we think we know that are actually wrong! So, for example, the teams might be asked to name an animal that buries its head in the sand. A contestant saying ostrich will have points deducted because no-one has ever seen an ostrich bury its head in the sand. And yet the myth dates back to Pliny the Elder. You get the idea of the show.

In a presentation at the Share user conference last Wednesday entitled “Hex, Lies and Videoblogs”, IBM’s chief architect for cloud computing, Frank DeGilio set about (in the style of QI) debunking some of those mainframe myths that we all come up against time after time.

For many years, the much-missed Xephon organization published the Dinosaur Myth, which crunched the figures for small, medium, and mainframe installations, looking at hardware, software, maintenance, and running costs. And every time it did the sums, it found that mainframes cost less overall. Similarly, DeGilio pointed out that most people ignored any figures apart from the basic hardware and software ones. He argued that, particularly for large-scale infrastructures, management complexity and personnel costs are often critically important parts of a system's final price tag. As Xephon’s publication identified, an expanding infrastructure requires more people if it’s distributed than if it is a mainframe.

There’s also the general ignorance that mainframes are your dad’s technology, and if the code isn’t written in Latin or Ancient Greek, then it’s the digital equivalent. While it’s perfectly true that mainframes run a lot of COBOL and Assembler programs, they have embraced modern trends as they have occurred over the past 50 years. This means, as Frank DeGilio pointed out, that J2EE, Linux and other modern open standards are all widely supported. Perhaps more importantly, Frank asserted that there's nothing outdated about the way mainframes handle workload management. In fact, their ability to fine-tune resource allocation based on application need is far more granular and sophisticated than that of most distributed systems.

Old iron tends to break rather than bend. But that’s not the case with mainframes, which are highly flexible and well able to balance workloads. As DeGilio says, “the very concept of capacity upgrade on demand was ‘pioneered’ by the mainframe”.

It doesn’t matter whether a computer doing nothing is a Raspberry Pi or the fastest supercomputer in the world – it’s still doing nothing! What really counts is how the box handles real-life mixed workloads. Various figures are produced suggesting that specific hardware and software combinations can set specific benchmarking records, but what’s needed is something that can handle everyday workloads as speedily as possible. DeGilio argued that the mainframe’s flexibility means that its speed in handling multiple real-world tasks is greater than what might be indicated by testing a box to perform a single activity.

Like Stephen Fry, who chairs QI, it’s really the job of mainframe professionals to go out there and debunk these myths. There’s a clear need to highlight incorrect thinking and identify ways that the mainframe could be a better answer to IT problems. QI isn’t dull – so you don’t have to be the most boring person in the world just because you want people to change their thinking and see the bigger picture!

And finally (from the QI stock of knowledge), where do Panama hats come from?
The answer is Ecuador!

BTW: if you get a chance, can you ‘Like’ iTech-Ed Ltd on Facebook. You can find it here. Thanks.

No comments: