• Systems Engineer Consultant -...
• Cisco Solution Architect
• Test Engineer - IP Network (M4)
• Application Delivery Manager
• Project Manager

Enterprise OSes in transition

Tim Fielden, Information Age

18/02/2001 17:11:41

Every modern IT department has to manage its client-side services. But not so long ago, that wasn't the case. Back then a business could let its operating system do what it did best - and that usually meant crunching numbers.

Of course, operating systems have since taken a dominant role in the assembly and implementation of business applications. The mainframe-and-tethered-device setups of yesteryear have given way to OSes at both the client and server ends.

And the OS market has been flooded with new entries, thanks to such developments as the Internet, the wireless connection, Java, and a plethora of devices that run on proprietary OSes (including handheld personal devices, Web-enabled cell phones, and smart cards). To complicate matters more, some old favorites, such as Unix-based systems, have staged dramatic comebacks.

The good news is that businesses have more options than ever when selecting an OS. But the bad news is that, with operating systems now a major determining factor of success or failure, the costs of selecting the wrong one can be astronomical. After all, nobody wants to be stuck with an underpowered OS when the next killer app comes along. And combining multiple OSes, while sometimes effective, can sometimes expose you to interoperability risks.

So how do you select the OS that's right for your business? In the IT world, newer is often thought to be better. But that's not always true of operating systems, which sometimes endure far beyond their expected lifespan. For example, the mainframe and MVS combination ruled the earth back in the 1960s and 70s - and many businesses, realising what a solid platform it is for hosting Web sites and running Linux, still use it (or a variant thereof).

The rearview mirror

Many old OSes survive because their manufacturers know how to adapt them to current trends. Take the ever-popular AS/400 server and its OS/400 operating system. IBM's architecture, which dates back to the late 1970s, was once thought of only as a report-writing machine, but it has since transformed into a platform capable of running any business from the mum-and-dad shop to the international conglomerate.

Like the mainframe, the AS/400 and OS/400 originally tethered users to dumb terminals. But as communications took off in the 80s, IBM responded with a PC-based user interface. And when critics argued that the platform was too closed and proprietary, the firm rolled out Java support and a set of Posix APIs that, to this day, power many Unix applications.

Speaking of Unix, the old Bell Labs OS is another prime example of an operating system with staying power. Unix has always been a mainstay for universities and businesses that need scalable processing power, but smaller shops tended to stay away (partly because installing Unix was too difficult for many businesses and partly because of binary incompatibilities). In short, Unix had to improve - and that's exactly what happened. Today, businesses can select from a dizzying array of Unix variants, including Sun's Solaris, IBM's AIX, or any number of Linux flavours. The new Unix-based OSes are less complicated, more conducive to productivity, and just as powerful as the Unix systems of old. But they're Unix-based all the same.

Why are these old OSes still with us? Partly it's because they were well designed in the first place. But sometimes businesses just don't have good reasons to switch to a newer OS - and they have plenty of reasons to stay with what they grew up on.

The most common reason, by far, is software compatibility. It's not unusual for companies to try shifting from one OS to another, only to find that changing infrastructures while maintaining their full roster of applications is harder than they ever imagined.

And then there's Windows, which supports so many business applications that it has become a de facto standard for many organisations.

Just as importantly, Windows signalled the dawn of a new age for the enterprise OS by merging graphical interfaces into the operating system itself, thereby minimising the differences between the server and client systems.

Thanks to its emphasis on graphics, ease of use, and multiple process support, Windows quickly found its way beyond the desktop and into the business corridor. Many shops began to run Windows at both the server and client ends. So great was Windows' influence, in fact, that IBM's OS/2 system - which many believed was a superior product due to its object-oriented nature, pre-emptive multitasking capability, and robust 32-bit architecture - was run into almost complete obscurity.

Granted, as the OS's acceptance grew, so, unfortunately, did its complexity. Yet modern enterprise computing was profoundly affected by Windows. After all, it was Windows that showed businesses that applications could be created at the desktop and scaled to the enterprise with few revisions. But Windows also taught businesses many hard lessons about software scalability.

When Windows 2000 launched, many shops found themselves unable to upgrade due to legacy issues with their hardware and software; others could not find the staffing and support needed to accommodate the new technology.

One of the hottest trends at the moment is the "mix and match" approach to OSes, in which multiple operating systems are united on the same platform. Few industry observers believe that the next few years will bring revolutionary breakthroughs in the OS market, but most agree that OSes that enable tighter integration of disparate systems are here to stay.

By blending OSes, applications that once ran on only one platform can now run on several. For example, OS emulators such as Win4Lin, WINE, and VMware enable users to store, run, and interact with Windows applications on Linux machines (see reviews of VMware and SuSE Linux 7.0 at www.infoworld.com).

It may not be an entirely new concept, but OSes and the applications that run on them have become more open, extensible, and ubiquitous. OS interoperability is in higher demand than ever before.

With that in mind, it's a safe bet that the OSes of the future will be based on open industry standards. But businesses will still demand functional stability and compatibility, operational stability, interoperability, and scalability. That's one thing that's not going to change.the bottomlineSelecting an enterprise OSBusiness Case: Today's enterprise requires a robust, scalable platform to power its business processes. Choosing an OS is a momentous decision, but thanks to integration and interoperability technologies, even a bad choice can often be turned into a workable solution.

Technology Case: Application support, both at present and in the future, should be a top priority when plotting an enterprise OS strategy. The Internet and various new device types have led to increasingly complex applications, making the OS decision more critical than ever.

Pros:

Wide range of choices available

Integration technologies enable mix-and-match OS management strategyCons:

Inappropriate OS selection can cripple almost all IT-related business processes