ViewYonder

Way back in 2006 I was doing a VMware consulting gig in the United States’ mid-west, transforming how an organization approached, deployed and operated VMware Infrastructure.  One of the outcomes of this work was the white paper, Roadmap to Virtual Infrastructure: Practical Implementation Strategies.

Central to the theme of that consulting gig, and the Roadmap paper, was the VMware Maturity Model (VMM): this was a five-layer model like the Carnegie-Mellon Capability Maturity Model (CMM), but specifically for VMware Infrastructure instead of software development.

The VMware Maturity Model

There are eight process areas in the VMM:

1. Provisioning VMware Infrastructure
2. Consuming VMware Infrastructure
3. Security
4. Operations
5. Change Management
6. Configuration Management
7. Incident & Problem Management
8. Capacity & Availability Management

For each of these process areas, you give each one a VMM score.  Something to understand about the VMM: you have to progress through each layer from 1 to 5: you can’t do 1, 2 then skip a few to 5.

1. Initial / Adhoc – this is where you are starting from scratch and doing the process activities “off the cuff” because there is no process in place.
2. Repeatable – you’ve been doing the process activities a few times and are now doing them consistently.
3. Defined – now you are repeating consistent processes because they are also documented.
4. Managed – with metrics in place, you have a management view of the processes.
5. Continuously Improving – acting upon the metrics, you are improving all the time but not acting off the cuff: only improving based on data evidence.

After a few weeks, we’d get a scoring matrix together that helps us write a roadmap and get things better:

Why the VMM is upside down

2006 was a long time ago.  The VMM hasn’t really changed since then, but something has always nagged me about it.  On paper it looks common sense, but I’ve often found that it is hard for organizations to get to VMM3 or any higher.  There is some kind of gravitational force holding people to VMM1 and VMM2.  I now think that instead of it being a gravitational force where 1 is the ground, I think that VMM5 should be the ground and the target / highest level is instead VMM2!!

Here’s my thinking:

Being successful with processes is about a culture of applying scientific method to IT: you write a hypothesis (a procedure, like a test plan procedure) – then you test it, and based on the results you either improve the procedure (re-write the hypothesis) or improve the tester (learn more about the procedure) – then test again.

This is just like the ITIL plan-do-check-act cycle, but according to the VMM you don’t do this kind of work until you are at VMM5…   it’s the wrong way around… instead of waiting to go through all of the other levels to get here (from initial and through repeatable, defined, managed), instead I think you have to start with this scientific approach otherwise you won’t have a culture of process and will never get anywhere because without that culture you will always be fighting fires, performing heroics, and documenting nothing.  Perhaps if this culture is the rocket fuel, then when it is missing it explains why people can’t get up the VMM?

So, instead I think the VMM should look like this:

1. Use scientific method to write-test procedures, including metrics which measure outcomes and timings.
2. Governance in place to promote (rewards) and enforce (discipline) culture of scientific method.
3. Functioning re-usable library of the procedures.
4. Have procedures that apply to all activities related to data center 3.0
5. Teach and learn from others outside your organization (see good2great, about relativity!)

Why bother with a VMM?

The IT Process Institute has done many studies on what makes a high-perfoming IT organization compared to the rest.  They have a very mature, academic and data-driven approach to this and they have proven that organizations with a culture of process, where they are efficient, effective and well governed, have a higher throughput with lower costs.

This means a high performing IT organization is delivering eight times as many projects with five nines availability: what $$ value is that?  What goodwill does that create around IT?

It means that staff are now problem solvers, continuously creating new ways to do work (process does not equal static activities!) and being happy in their new found creativity: this attracts top IT talent, who have been shown not just to value $$ but also working for a progressive organization that values and recognizes quality of work amongst peers.

Controls are not barriers, but enablers: change management is like the brakes on a car, allowing you to go faster (quote from Visible Ops).  This can translate into reduced investment in tools because you are no longer blindly applying technology and instead you work with specific vendors to get exactly what you want.

The VMM is dead; long live the VMM!

I plan to work on this more as part of my VMworld 2009 session: High Performance vSphere Operations.  I’m putting the fruits of my efforts into this blog, under the Operations section, starting with the Data Center 3.0 Scientific Method discussion.