Dave Welch, CTO and Chief Evangelist
As we move forward, there are more layers in the system stack. Virtualization is being introduced for a whole bunch of reasons, each of which is justification alone in my mind for doing so.
From my viewpoint, virtualization blends the lines somewhat between the various IT administrators’ functions. I see that as a good thing for the admins, for IT, and for the business overall. This could be a huge paradigm shift in particular for some shops we’ve worked in where the various IT teams barely communicate between themselves.
My first-steps formula for moving in this direction:
- Select up to ten performance attributes from these hardware resource groups that I list in what is usually descending order of importance: IO, memory, CPU, and network.
- Adopt metrics for each attribute.
- Adopt a handful of performance tools freely available or bundled with native Linux or Windows.
- Author and promote elevator-pitch summary conceptual overviews of the performance attributes, the needs they fulfill, and your associated SLAs.
- Regularly publish your performance metrics to the organization including the C-level and business users.
I believe there is significant benefit to IT proactively educating superiors and business peers on these performance attributes, unilaterally suggesting associated SLAs/red lines, and freely making the information available. Rather than self-inducing pressure, such information liberates. It naturally removes organizational barriers. It goes a long way to removing fear, uncertainty, and doubt. When appropriate, it brings more focus and leverage to IT CAPEX discussions including budget exceptions when necessary. Last but not least, it can demonstrate that VMware Infrastructure is for all intents and purposes transparent.
I’ll publish some key performance attributes, straw-man metrics, and suggest a handful of tools in a follow-up blog post.