In this post I examined the fundamentally different assumptions that underlie different forms of virtualization. In particular, it is precisely the assumption that we can use commodity infrastructure safely that leads to the tremendous deployment and operational cost reductions, as well as the equally impressive scale and reliability improvements delivered by application fabrics.
Now a quick thought as to how this applies to service virtualization.
Some vendors are touting “service virtualization” to mean that service definitions and composite applications may be defined (independent of the particular platforms) across various hardware and software platforms. That is good so far as it goes, but still leaves individual services dependent upon individual computers (servers) for their sense of scale, reliability, and so forth. In other words, if a service foo really needs to scale quickly you’ll need to scramble to move other services off, look for a larger server to put foo on, or both. Not to mention that you’ll still need to code foo to be able to survive infrastructure failures.
A certain improvement, to be sure, but much more is possible.
Suppose those service definitions point to application fabrics rather than specific, traditional servers?
That would be the best of both worlds – a platform independent framework for defining individual services and their interactions, combined with a commodity infrastructure that could scale (or shrink) as needed, be as simple to operate as a single computer, and never break.
Now that would be real service virtualization!
Technorati Tags: application virtualization, commoditization, service virtualization, virtualization
Comments on this entry are closed.