Virtual machine technologies hold the promise of unlocking operating systems and application instances from dedicated hardware hosts. They allow those hardware hosts to house many operating systems instances (the average is between four and six in our experience) with applications, concurrently, yet largely autonomously. More bang for your hardware buck is always good news.

The news just gets better with the recent advent of virtual instruction-enabled CPUs. These processors, which have been altered to render fast support to CPU instructions that are required to fully isolate virtual machines, play a large part in a single server’s ability to house several concurrent instances of operating systems efficiently. Sixty-four bit CPU architectures also present a seemingly limitless memory map for all of the ‘beings’ to live inside a single server. The upper limit for memory on 32-bit architectures has typically been 4GB of memory, and increasingly, this represents application servicing constraints as additional requests must be paged in and out of 32-bit confined segments. Virtual machines made from 32-bit memory model applications and operating systems map easily and generously into a 64-bit memory model, taking but a fraction of the 64-bit model’s turf.

Take the dual-core servers popular in corporate implementations today (and not to mention the three-, four- and even more-cored machines on the foreseeable horizon), and a single server ‘box’ now has resources available to power numerous multiple instances of autonomous processes easily. What “multiple” might mean to your environment is an exercise for only you to compute based on your operating system choice, your application requirements, and both your risk tolerance for a single point of failure.

(Compare physical and virtual server management tools in our revamped Server Management Buyer's Guide.)

< Return to main test