In a blog post yesterday, Cameron Haight writes about “The Need for IT I&O Reform”. In it, he addresses a critical issue: the clash between existing manual operations processes and
the need for agile, responsive, automated cloud operations processes. The money quote:
“At the end of the day, I believe that it is no longer tenable to simply layer an increasingly complex operations management tool, organization and
process infrastructure on top of an already complex software and hardware foundation and to expect satisfying results.”
[For timely cloud computing news and expert analysis, see CIO.com’s Cloud
Computing Drilldown section. ]
He particularly addresses ITIL, which is a highly structured set of processes designed to standardize IT operations, and states that it may overly
focus on placing limits on flexibility, when a potentially better approach might have been to focus on the insights of those at the coal face — the sys
admins, operators, console operators, and the like. Based on their insights and experience, IT I&O could be more responsive and more efficient.
Haight notes that there is a growing “movement” addressing the need to make IT I&O more efficient. This movement goes by names like devops
(which I wrote about a couple of weeks ago), agile operations, agile infrastructure, etc. This movement is in contrast to a dumbed-down approach to
I&O, which seeks predictability by constraining options.
I think Haight is absolutely right that the current approach to I&O is unsustainable. The highly-manual plethora of “Bs” (the Configuration
Management Data Base, the Change Control Board, the Incident Tracking Board) acts as an obstacle to agility and
innovation. And when you throw in scale (a hobbyhorse of mind that I ride in
another blog post), what is merely inconvenient becomes untenable and unmanageable.
I’m not sure I align with a perspective that advocates letting those at the coal face make the decisions. I say this for several reasons:
• Many organizations do not hire skilled “coal miners.” That is to say, there are only so many smart, gifted, experienced I&O people to go around.
Most organizations won’t have the sharpest knives in the drawer. And the solutions the latter implement for administration are poorly designed, buggily
implemented, and probably as poorly suited to scale as the manual alternatives. Standardized processes are usually implemented not to prevent
brilliance, but to prevent the effects of stupidity. In effect, they replace high variability of outcomes with predictable mediocrity, which enables more sleep
at night for senior management.
• One person’s skilled “coal miner” is another’s undisciplined hacker. The elegantly architected system administration edifice developed by one
smart person becomes an unmaintainable mess when inherited by another smart person. Everyone’s had the experience of a really smart person leaving
an organization and no one else being able to make heads nor tails of what he leaves behind. I recently had a conversation with a highly-talented engineer
at a company and discussed with him the use of one of the commercial management frameworks for Amazon Web Services. He pooh-poohed the use
of a framework with the comment “I would just write one myself.” Easier for him — more difficult for everyone else down the line.
• It’s inevitably idiosyncratic rather than standardized. Even when a number of people collaborate, the output, sans a standard approach, is a
one-off, which makes bringing new team members up to speed difficult and imposes a high operational cost on the I&O operation.
A better approach is to identify the emerging best practices from the devop movement and standardize on those. The fledgling products Chef and
Puppet are steps in the right direction. Organizations trying to break free of the “process” box should look to products and practices that will be widely
used across the industry, thereby avoiding the “IH” (invented here) syndrome, in which an organization charts its own course only to find itself lodging at
a high-cost hotel.
I may be misreading Haight’s message — he notes that he will be doing additional research in this area — and he’s not suggesting ad
hoc processes in place of clumsy manual ones. If I’ve over-read into his post, I apologize, because I think it focuses on an extremely important issue.
The key thing is that whatever emerges to replace languid manual processes must be automated, standardized, and support very high scalability. What
we really need is a way to identify and standardize the inventions of really smart people who are today solving the inevitable future problem of scale and
agility — and to apply their solutions everywhere. That’s how IT I&O will be reformed.
Bernard Golden is CEO of consulting firm HyperStratus, which specializes in
virtualization, cloud computing and related issues. He is also the author of “Virtualization for Dummies,” the best-selling book on virtualization to
Follow Bernard Golden on Twitter @bernardgolden. Follow everything
from CIO.com on Twitter @CIOonline