When CIOs are procuring new IT systems, they naturally look for those that are the most cost/effective. To put it another way, they aim to find what best meets their business requirements for the lowest price.
So far, so obvious. But how do we make the most cost/effective choice? For a number of years –roughly from the late 1980s until recently – the most favoured option was what I’ll call the Multi-Source Model. The hardware and software are obtained from a number of suppliers, each of which is regarded as the best or most cost/effective option.
The model dominated thinking for a long time because of its assumed benefits. In particular, a choice of software components such as database managers and application servers led to the view that costs would be lower. There is, however, a hidden assumption – perhaps wishful thinking is a better way of putting it: all the products would come together and work without any significant problems.
Experience has shown that this optimism is misplaced. Problems have emerged in two broad areas. First, the task of integration has proved far harder – and of course far more expensive – than was anticipated. Getting a variety of different products to work together is not simple. And when a new release of a product appears, at the very least a new round of testing is required. Release cycles don’t help either. Different companies produce products on different cycles, and some releases are not compatible with other products.
The second group of issues is to do with optimisation. In the multi-source approach, there is no real scope for optimisation by moving functions between software components, or even into hardware, as each component has a different owner. Such optimisation can greatly help performance, recovery, resilience and security, which really do matter in mission-critical environments.
The Integrated Stack Model, an approach long used for ClearPath systems, solves both of the groups of problems. Start with integration. The hardware and software are designed, developed, integrated and tested by Unisys before release, minimising or eliminating clients’ effort for integration and testing. Compatibility issues between different components do not arise as there is only one source for the products. And there is just one port of call for support, eliminating the need to identify the precise location of any problems and avoiding finger-pointing between different vendors.
Supplying all the components also allows Unisys considerable scope for optimisation. Functions can be moved between software components. Special functions can be implemented in hardware. And specialised interfaces between components, based on a hardware/software combination, can be developed. These opportunities have been exploited to the full, providing the high levels of security, availability and performance we see today in ClearPath systems.
Others are now jumping on the bandwagon. Erstwhile followers of the multi-source approach have now recognised the value of the integrated stack. The most high profile example so far is Oracle, which has adopted ‘Hardware and software: engineered to work together’ as its marketing strap line. In spite of its claims, Oracle still has some way to go to integrate its diverse range of products into a coherent single stack. Unisys is already there.
I go into the subject in a lot more detail in a white paper titled, Unisys ClearPath Systems: Integrated hardware/software stacks.