Restart

For those few who have followed my blog with earnest through the first half of 2009 and have noticed that I haven’t updated this space for a while please consider this a restart. I am again committing myself to update this space on a weekly basis with more of my technological observations.

First an update -- and full disclosure -- I am now VP of Marketing for a small virtualization company called ScaleMP. ScaleMP’s own technology would easily be a natural fit within this blog but I needed to make sure the full disclosure is present before going into further detail. I do have an interest in this technology -- both philosophical and financial -- so continue reading with that warning in mind.

Where all other virtualization vendors -- VMware, Microsoft, Citrix, Red Hat among others -- have focused on using hypervisor / Virtual Machine Monitors to disaggregate server hardware (or desktop hardware) from operating systems in order to increase workloads on top of a single system -- also known as partitioning a server -- ScaleMP has a hypervisor / VMM that aggregates multiple servers into a single system with a single operating system. This may seem counterintuitive but storage vendors have been both partitioning and aggregating storage systems for decades. On one hand -- and as a parallel to what VMware, et. al does for a server -- customers have purchased large storage systems and partitioned them for multiple users and multiple purposes, customers have also purchased a bunch of small disk arrays (JBOD for instance) and aggregated them to appear as a much larger disk. ScaleMP’s technology allows customers to purchase far less expensive x86 servers and aggregate them together to create a large supercomputer with hundreds of CPU cores and large RAM / memory capabilities. These systems appeal to anyone with a high performance workload needing many CPUs or working with files in excess of 128GB but don’t want to pay millions of dollars for the system capabilities. Using off-the-shelf x86 servers and ScaleMP’s vSMP Foundation aggregation platform can now assemble a 128 core, 4TB system running Linux for a few hundred thousand dollars.

As virtualization for consolidation / partitioning was revolutionary when IBM introduced it over 25 years ago and took flight when VMware introduced the capability for standard Intel servers I believe the same will be true for virtualization for aggregation. The challenge for server vendors is realizing that forcing customers to rely on proprietary, expensive high end systems for exclusive workloads is counter intuitive and flies in the face of technological evolution.

This blog will not exclusively focus on a single technology -- certainly with Apple’s iPad tablet announcement looming mid-next week I’m sure that will result in commentary here in the not too distant future -- it is absolutely a certainty that virtualization will continue to be an underlying theme of many of my musings.