Tuesday, January 11, 2005

fyi Novell, Red Hat Eye Virtualization for Linux, OS rivals to offer open-source tools


Pointer to article:

Kobielus kommentary:
Open sourcing of virtualization technology makes good sense. By definition, virtualization is decoupling of the interface from the app, the app from the OS, and the OS from the underlying hardware platform. Virtualization enables flexible, logical resource partitioning and/or aggregation across diverse platforms. True platform independence. Hardware vendors want to tie you to their metal, OS vendors to their environments, and app vendors to their code. None of them has a real interest in decoupling you from all of that legacy.

I suspect that Xen and other open-source virtualization projects will achieve the same critical deployed mass that catapulted Linux, Apache, and other seminal distributions to widespread adoption (though Xen’s licensing-related lack of support for Windows shows how the platform vendors can throw a monkey wrench in the open-source virtualization game). I don’t doubt that Microsoft, VMware, and other virtualization software vendors will do a good business. But only an open-source community can muster the diverse resources necessary to sustainably support the “virtualization software to run [dozens of] guest operating systems [and several times as many applications] on [growing numbers of commodity] production server [chassis], each with [many physical blade servers, each of which has more than a handful of] CPUs [and each of those is logically partitioned into myriad virtual machines]” scenarios, such as in this article.

The advance of virtualization will spell the end of forced OS and app migrations. The legacy can live forever if we can physically/logically slice-and-dice new hardware to run old software. And let's not forget grid technology: a way of aggregating old and commodity hardware into larger environments that transcend any OS/app/server. Call that the bunch-and-crunch virtualization scenario.

Virtualization technologies let old data-processing assets live on as long as they can carry their own weight, in terms of business application.