SANs And Software Development - Storage Area Networks - Technology Information

Computer Technology Review, Dec, 1999 by Fred Moore

The Storage Area Network is rising in prominence and promises to redefine the model of dedicated storage connected to a server. SANs are separate networks of storage devices dedicated to managing, storing, and retrieving data without tying up the computing resources of the server. The operative word here is managing and managing data is clearly based on software. The storage industry has demonstrated that it can significantly and continually improve the devices that store and retrieve data in a timely manner. The computer industry has not demonstrated that it has found a comparable path of progress for developing software. To make SANs fulfill their promise, much of the remaining functionality will come from software serving as a type of over-arching operating system for SAN management. Today, software technology that effectively utilizes and manages SANs is relatively immature. On a broader scale, software is rapidly becoming one of the key enabling elements for the overall IT (Information Technology) indust ry. Let's examine some of the issues presently affecting timely software development.

Software development has been the target of numerous tools and techniques over the years, but few of these have had much impact on software productivity. Today, the progress of the global IT industry is almost entirely dependent on software programs. Most industries are dependent to a greater or lesser extent on IT. Without software, hardware is unable to function. It would seem natural to suppose that business and governments would assign a high priority to determining how to write reliable software in a timely, cost-effective way. This has not been the case and progress in developing software has been slow. Theoretical concepts like structured programming, object-oriented programming, databases, and other formal methods have gradually found their way through the industry until some of them have become quite widely understood several decades after their initial appearance.

When investigating the likely advances in software development beyond the year 2000, we must be aware that there already exists a collection of well-documented programs, processes, and techniques, most of which are not used in the average development project or even known to the team members. Long-term trends like Moore's Law and its analogues for storage and bandwidth improvements have steadily pushed back the boundary of the impossible. We can be confident that these advancements in processor speed and storage capacity will continue for the foreseeable future. Computing power and storage capacity are not presently limiting factors for the Information Age; improving software development is, however. We have no Moore's law for software improvement. The consecutive invention of assemblers, compilers, editors, third and fourth generation languages, structured languages, Object-Oriented Programming, rules-based systems, CASE, repositories, databases, and data warehouses have advanced much more slowly than their hardware counterparts.

The benefits of systematic software re-use have been appreciated for a long time. Technical considerations apart, the introduction of software re-use has encountered the human characteristic called "resistance to change." Because of this resistance, average software development practice falls far short of what is theoretically possible. The Information Age will be held back without fundamental changes occurring in software development. Looking a few years ahead, we can imagine a more sophisticated developer's workbench taking advantage of the existence of plentiful libraries of re-usable components. In the interest of quality and reduced development cycles, as much of the development process should be automated as possible.

In the future, software developers will assemble components into finished systems, much as server manufacturers assemble finished systems from chips and motherboards. They will have much more time and energy to spend on understanding requirements and making sure the end users are satisfied. The inherent flexibility of a component-based system will make it much easier and quicker to change and upgrade software as business needs evolve. We have examined numerous predictable and encouraging trends of progress in the storage networking industry in the past except for software development; now it is time for software development to join the progression.