Monday, August 28, 2006

Gartner Prescribes Software Appliances to Relieve Windows Bloating

In a recent report covered in techworld, Gartner analysts Brian Gammage, Michael Silver and David Mitchell Smith declared:

""The current, integrated architecture of Microsoft Windows is unsustainable - for enterprises and for Microsoft."

Said another way:

"The era of the bloated, 'one size fits all' operating system must come to an end."

The article goes on to explain how virtualization and software appliances are the future of the enterprise operating system, even for the desktop. Drivers and basic system services will be provided by a streamlined OS that is tightly integrated with the hardware. New applications will arrive as software appliances, with their own tailored operating systems. These software appliances will integrate with the hardware OS via a virtualization layer such as VMware, eliminating all of the conflicts that arise when applications that install natively on the hardware OS begin competing for OS system resources. With software appliances, applications will truly "plug and play."

According to the article:

"Microsoft doesn't agree with this vision, saying it's identified problems with integrating data across partitions and creating a consistent user experience."

Of course they "don't agree with this vision" because it eliminates the innovation stranglehold of the standard operating system that perpetuates Microsoft's monopoly power. The software appliances that carry the value of the application don't even have to run Windows because the virtualization layer makes the OS irrelevant. The power of Microsoft is their abilty to charge more money for their applications because they own ALL the layers from the hardware up through the user interface. When the operating system functions get split between a hardware OS that provides drivers and system management and an application OS that provides system services such as file systems and libraries, the stranglehold is broken.

You may still need a Windows application OS to run Office, but if Microsoft does not tailor it to simply support Office, it will not be competitive. The footprint will be 10X what is required to run Office, and customers will balk at this gross inefficiency foisted on them by Microsoft's unwillingness to abandon the practices that were once so profitable but now are impractical given the changes brought on by virtualization. Customers always win . . . . . eventually.

Sunday, August 20, 2006

Ingres beats Oracle to Linux

Matt Marshall (no relation) of the San Jose Mercury News sounded the alarm this past week at Linuxworld, declaring that Oracle was going to ship their products pre-configured with a Linux operating system based on Red Hat's product. It seems that Matt's prognostications were a little premature. The biggest news on the floor actually came from an Oracle rival, when Ingres announced Project Icebreaker. It seems that while Larry Ellison has been talking about Linux, Ingres has been working on their own software appliance, which is currently in beta and will ship in a few months.

I fully expect Oracle, and almost every other software vendor for that matter, to eventually ship their products pre-configured with a Linux operating system. The benefits of the software appliance approach to both customers and the software vendors themselves are indisputable, and the open source license of Linux makes it a perfect OEM component for the software vendors. As virtualization becomes pervasive with Intel VT and AMD Pacifica, the argument for the standard, general purpose OS is going to get weaker and weaker.

Although I don't doubt that Oracle will someday embrace software appliances, I would be suprised to see them ship their products pre-bundled with the Red Hat distribution. Red Hat does a great deal of work to create and maintain a general purpose operating system. Because of the "one size fits all approach" of a general purpose operating system, an extraordinary number of patches and components are included in order to serve the broadest possible needs of the market. If Oracle chooses Red Hat Linux as the basis of their software appliances, then they will be forced to maintain a much larger footprint for Linux than their products actually require. For a company that seems obsessed with operating efficiency, adopting an inefficient approach to providing software appliances seems unintuitive.

I guess we will all wait with baited breath for the news from Oracle Openworld in October. . .

Sunday, August 06, 2006

TCO with a Baseball Bat

A recent Forbes cover story profiling Larry Ellison and Oracle proclaimed:

"For every dollar corporate customers spend on new software, they spend $6 implementing it and getting it to talk to other programs."

Put another way:

"If you think it costs a lot to buy a license for software, wait until you see how much it is going to cost you to use it."

What a terrible indictment of the software industry. Even the open source vendors who clang and bang about the expense of proprietary software licenses are often missing the big picture of the user experience. And customers are complicit in this debacle when they accept that they are going to have 4 levels of server infrastructure associated with promoting software: development, test, QA, and production. Why did you pay so much money for a "certification" if nothing is really ever certified until you, the customer, put it through its paces?

It is not surprising that SaaS vendors such as and appliance vendors such as Barracuda Networks are now the rage of the software industry. Both of these companies assume responsiblity for component integration, and they build their software to interoperate with every other required software element via standard protocols like TCP/IP. It is also not suprising that VMware is growing 73% annually by providing technology that enables virtual appliances, decoupling the infrastructure layer from the application layer which reduces "certification" to copying a file to a system.

I still chuckle every time I read a Microsoft funded study that reveals the TCO benefits of Windows server relative to Linux. In most cases, Windows is proclaimed to be anywhere from 10 - 25% cheaper to own for some certain workload - the equivalent of counting gnats on a rhino's ass. The point of Linux from a TCO perspective is that for Unix based applications (which won't run on Microsoft Windows anyway), the customer gets 3X the application performance on hardware that costs half to one third the price of proprietary Unix, yielding a price performance benefit of 6 - 9X. This breakthru performance for the customer is TCO that swings like a baseball bat.

Like Linux in the Unix market, SaaS, appliances, software appliances, and virtualization all swing TCO like a baseball bat. If you are a software vendor, whether open source or not, you better start warming up in the on-deck circle with a similar capability, or you will soon be headed for the showers.