We all know that virtualization is a fast-growing trend. Various studies and analyst reports (Yankee Group, Forrester, Gartner) show there is a clear set of operational benefits from virtualising physical servers, and the vast majority of organizations intend to continue or accelerate this trend.
No problem – we can keep deploying tools such as VMware to increase the number of virtual machines, keep consolidating the server infrastructure, keep whittling down the environmental & facility costs, and scale up the IT operational support. Right?
Wrong. Most reports give an average ratio of virtual-to-physical as around 30-40%, and that to move past this level requires considerably more effort and management. Forrester’s understanding is deep enough to define a 4-stage maturity cycle, with stage 3 being process improvement. (ref. “Virtualization Management and Trends” Jan 28th 2010).
What is this management effort and why is it needed?
I’ll answer the why part first. What the studies show is that for continued virtualisation growth, IT capability has to scale up and this requires processes to be improved. The 30-40% threshold appears to be where the existing and largely manual operations processes start to limit the ability of IT service operations. (A good study on this topic is from Forrester’s Glenn O’Donnell, published in July 2009. Forrester’s Jan ’10 paper noted above on virtualization trends is also very good.)
So this observation exposes an assumption in my earlier statement “scale up the IT operational support”. That simply doesn’t work. Managing an increasingly virtualised infrastructure means change and that means building a process capability. The typical early stages of virtualising rely on subject-matter experts learning the technology and the best ways to optimize usage, crafted around their organisation’s needs and priorities.
Building operational capability also means going through an organizational change so people can work effectively with new processes and tools. This is well-understood and has been for many years. Whatever discipline is adopted – ITIL/ITSM, LeanIT, SixSigma among several – the path to improved operational efficiency requires, and always has, standards and disciplines to be developed then absorbed into BAU.
Now for the what part – what is the extra effort required? What processes and tools can be applied? If we think in terms of outcomes, a few that shape the trend: 1) standardize on IT capabilities like resource scheduling & policy compliance ; 2) self-service with provisioning; 3) service delivery orchestration; 4) moving to centralize servers and network components.
To re-cap: To continue with virtualisation benefits requires a scale-up of operational capability. To achieve that needs process automation, a well-understood discipline.
You might think automation means less people doing more work, or less-skilled people replacing skilled technicians and systems administrators. Automation is a dumb-down.
No, quite the opposite. One of the drivers for automation, as reported in various analyst studies, is that increased virtual deployment can create a skill shortage. We need automation to enable skilled administrators to get more done, deploy and configure more virtual systems, and manage a larger-scale virtual infrastructure. There is no appetite in the corporate IT spend to increase headcount for these efficiencies, and even if there was, the processes that work for managing 35% of servers virtualized will not scale to manage 80%, 70% or for most organizations, even 50%.
A sample of the kinds of tasks that need automation for scalability:
• standardized policy-based configuration and detection
• provisioning, resourcing and de-comissioning
• capacity and usage management
While aspects of the “what” are specific to managing a virtual environment, the underlying “why” isn’t. Lack of process automation leads to increased manual effort, extra admin time handling operational problems, and becomes a barrier to scaling up the benefits.
The automation job is clear: optimise processes; standardise configurations; unify your technical silos; centralise services; and deploy tools that help enforce new processes. Simply relying on tools does not enable process efficiency.
Never mind the subject matter, this is Process Automation 101.
David Gandar
Managing Director
<< News < 2010 – the year of truth for IT Operations | IT Operations Problems Need Automated Solutions >
post a comment
Post Comment