Feature:

Natural selection

23 March 2007

Computers have come a long way since the mainframe. Spyros Sakellariadis tells Jacqui Griffiths why Microsoft must now be considered as a natural choice for enabling business growth.

Times change, people move on, says the old adage. What this means is that, as the world changes, those living in that world must adapt and evolve. Dinosaurs must have at one point fitted perfectly into their environment, before their world became unable to sustain them and new species evolved. A similar evolution has been taking place within the business world, where there is a pressing need to adapt to an ever faster, more competitive and global business environment. Information technology (IT) has played a major role in the development of this environment, but is also enabling organisations to adapt and survive in it. The mainframe exemplifies this - a large, lumbering and omnipresent species, that is shrinking as the majority of its members are unable to survive in the new world. For Spyros Sakellariadis, director of Microsoft's Mainframe and Midrange Alliance Programmes, it's time for businesses to move on. "The mainframe is a very well tuned device," he says. "It's been there for 30 years or so and it does its job incredibly well - it hums along in the background and it has many years of data and business intelligence encapsulated in it, along with 30 years' worth of debugging for the applications and so on. But we've come a long way during those few decades.

"While mainframe hardware and programming have improved dramatically over the years, not all of the programs have - some of the code running today was written in the 1960s. It's great stuff, but technology has developed far further than that and you have to ask yourself: is this a good choice for not only your back-office, but also for your most mission-critical line-of-business applications?" In this environment, it is perhaps no surprise that Windows has emerged as a key competitor to the mainframe. Microsoft's extensive partner ecosystem and investment in research and development have helped to establish Windows as the key business support system. But while the mainframe presents an imposing physical presence that speaks of dependability if also of antiquity, it can be difficult to grasp the benefits of the more nimble and individualised Windows environment as a viable alternative.

This is precisely because Windows has become ubiquitous in both business and domestic settings. "The very raison d'etre of the mainframe - the need for a highly scalable, highly reliable platform - is now attainable with Windows," observes Sakellariadis. "If you have a problem to solve, you have to look at all the different options. The problems that the mainframe was there to solve can be solved by Windows at a fraction of the cost of the other alternatives - so it has to be a consideration." Evidence of this is fast emerging. "In the US, for example, we see 911 (emergency response) systems running on Windows," adds Sakellariadis. "Worldwide, we see stock exchanges running on Windows, among them the London Stock Exchange, the Brazilian Stock Exchange, Euronext and the Madrid Stock Exchange. What better examples are there of whether Windows is there for the enterprise?" So, let's get down to the specifics. How can a Windows environment address the issues facing today's chief information officer (CIO)? "The CIO has operational challenges to face every day, whether it's cost reduction, a technology initiative, business change like a merger with another organisation, or standards, compliance etc. Those are tactical challenges." Sakellariadis comments. "Long-term, he's got a much bigger strategic issue - how to move IT away from being a purely support function that is a drain on the business. How does he transform IT from a piece of overhead into a strategic asset helps the business attract and retain more customers, supports innovation and makes the business an acquisition target because people want his technology?"

"Looking at hardware alone, the cost of the mainframe has traditionally been amortised over several years," says Sakellariadis. "This makes it something of an invisible cost. At the high end, mainframes are incredibly expensive, and the challenge is to load as many workloads on them as possible to keep the cost per application low. At the low end, less expensive mainframes have been introduced, trying to get them to compete well with the typical distributed systems running fewer workloads. But as Sakellariadis points out, these low-end mainframes are still much pricier than the equivalent Windows systems: "Even including discounts, the cost of a low-end IBM z9 BC mainframe with minimal memory, storage, and networking components may run you around US$135,000 (£105,000). Compare this, for example, to an HP Proliant DL 580 G4 with four dual core processors - that machine is much more powerful than the z9 BC, but can be bought for US$15,000 (£12,000), excluding discounts. If you want to upgrade with eight gigabytes of RAM, that will cost US$90,000 (£70,000) for the mainframe machine versus US$2,000 (£1,500) for the HP DL 580. So if we're talking about opportunities for reducing costs, really it's a case of commodity hardware versus sole-source vendor."

When it comes to software, aggressive pricing has seen large reductions in the cost of operating system licensing, from about US$2,000 per MIP (millions of instructions per second) a few years ago to about US$500 (£387) per MIP today. However, even at US$500 per MIP, this equates to US$1 million (£774,000) for a 2,000 MIPS machine. Meanwhile, a Windows licence may cost only around US$32,000 (£25,000) for a 16 CPU machine that can happily deliver 2,000 MIPS. According to analysts working in the mainframe field, the average mainframe shop pays about US$4,000 (£3,000) per year for software licence costs per mainframe application developer. On the Windows side, the same set of programming, debugging and testing tools might cost a quarter of that. "If you have 100 developers on staff, that starts to make a difference," comments Sakellariadis. "In addition, Visual Studio will improve the productivity of developers - if your developers are 20 per cent more productive, that's another 20 per cent reduction in your labour cost."

In the case of Stockholmshem, one of Sweden's largest housing companies, these savings were even greater. Its migration resulted a 50 per cent rise in developer productivity following only a week of training Visual Studio, SQL Server and the .NET Framework. Supportability is also a key issue - can you keep your systems running at all times? "One issue often raised by clients is concern about being able to find people given the greying mainframe workforce," says Sakellariadis. "There is definitely a people shortage in the mainframe world. It's not a critical shortage yet, but the average age of mainframe shop employees is increasing, and the new pipeline is very small. Meanwhile, the Windows pool is growing - it's at the level where it's a commodity. Microsoft makes people available and gives you a choice of every single supplier in the stack, whether it's hardware, applications or middleware. It gives you choice."

This was a key reason behind Stockholmshem's migration from the mainframe ? the average age of support personnel for its system was 55 years. As Stockohmshem information technology program manager Allan Hansson put it: "Cobol isn't the technology that younger people are flocking to." For many research firms at the moment, agility is quite the thing - it seems that companies that cannot be described as 'agile' are not worth discussing. And although a Windows environment is supposed to enable that agility, the term is rarely defined, an issue that Sakellariadis finds frustrating: "The simplest way to define agility is in terms of reduced time to market for applications - increasing the ability to react quickly to changing business requirements," he explains. "If you wanted your company to be able to get products out more quickly, the first thing you'd need would be to focus your programmers on strategic tasks rather than routine ones like writing and maintaining your own non-differentiating applications. The first thing you could do to increase agility is free up your programmer staff, get them off coding things like an HR system and buy a commercial, off-the-shelf one on Windows. The second thing is to improve the productivity of your programmers.

"In the Windows environment, your ability quickly to bring something out, test it, change it and so on is much, much greater than in a mainframe environment. It empowers the developer, as opposed to making them a small cog in an inflexible machine in the mainframe environment."

When it comes to making the IT system a strategic asset, the first question is: how can the system help to attract new customers and retain them?

"Most users now spend all their time in front of a Microsoft Office screen," notes Sakellariadis. "The future will see data embedded in Office applications, not custom applications. Look at SAP Duet ? the SAP data is exposed inside Outlook. If the average office user spends 80 per cent of their time looking at Outlook, that SAP data had better be available there, otherwise he's not going to look at it. Cobol applications on the mainframe are not that highly integrated with the user's environment at the moment." The second key strategic goal is driving innovation. "Innovation alone is not agility," says Sakellariadis. "Agility is working out how to implement an idea but innovation is fostering an environment that lets people come up with the idea in the first place. It's about collaborative work, finding ways to access data everywhere ? internet searches, desktop searches - new ways to capture, classify, share, search and find structured or unstructured data; new ways to analyse and detect trends you hadn't noticed before. These are very abstract concepts and tools. There's no way to force innovation, only ways to sponsor a climate in which new ideas can develop. A big strategic value of Windows versus the mainframe lies in fostering innovation. The mainframe is very good at repeatedly doing the same old task, but doesn't promote collaboration and innovative thinking.

"The top priorities are, develop new customer relationships, retain customers, be agile, be innovative. All that can only be done when your core business and data are off the mainframe."

The case studies referred to in this article, and more information about migrating to Windows, can be found at http://mainframemigration.org/default.aspx