There are a lot of things that are simple in principle that are a lot more
complicated in reality. International relations. Poverty. The personal
problems of other people (who stubbornly refuse to take your advice about
ditching that inappropriate spouse, or going to get a real job, or not
voting for that dangerous lunatic who’s clearly not qualified for an
Somehow the result is always stickier and more complicated than the original situation, though, no matter how clear your instructions were.
The same rule applies to the process of server virtualization—which is more
complex that it looks, but not nearly the tar pit of nearly any internecine
discussion that includes the phrase “when are you going to quit
pretending” or “and grow up.”
The principle of virtualization is simple enough when you explain it to the CEO and CFO:
Run more than one server on one server, save money on hardware,; cut back on power consumption; loosen restrictions on the kinds of hardware you can use; make disaster recovery a lot easier and cheaper; reallocate computing resources according to actual demand, not predictions based more on the CFO’s need for budget projections than the need of your colleagues for IT resources.
Of course, once the suits understand the principle and potential of virtualization, they have their own priorities:
Reduce the number of servers without buying bigger, more powerful servers first; save even more money on hardware; cut IT’s budget for power and A/C; loosen the requirements Purchasing uses to buy servers and save even more money on hardware; cut IT’s budget for disaster recovery; cut the part of the IT budget that’s based on covering peak demand for bandwidth, compute cycles or software licenses; cut IT’s budget for data-center space and support; cut IT’s budget for staff to support all that IT they’ve already cut.
The rule of unintended consequences. Of course, it’s not so much a rule as a rueful—a grudging admittance that in human relations, as in nature, any action has an equal and opposite reaction, but not in a form you’d always like.
Intelligence services call it blowback, and hate it for making descriptions of their work sound oxymoronic.
Janco Associates is already predicting that IT salaries will flatten or decline this year. IT hiring is at its lowest level since 2004 and a lot of freezes are already in place, according to Janco’s most recent annual IT salary survey. Raises in IT have been lower than the cost of living (which, if CoL includes gas prices, isn’t surprising).
Many companies, Janco’s survey says, have also cut or eliminated discretionary IT budgets, pulled hiring levels for even high-demand specialties like security and virtualization back to normal levels, reduced top-IT-executive raises to something like one-and-a-half percent, and started trimming lower-level and administrative jobs.
Janco blames the recession, which undoubtedly contributes. But the same pattern also makes perfect sense in the context of the other cost savings that stem from virtualization.
Do you really need the same number (or same salary level) of admins to manage 1000 virtual servers on 200 physical servers as you do to manage 1000 actual servers? Probably not, when a good part of physical-server administration is going to make sure a non-responsive server in a branch office is still plugged in.
APC wouldn’t say so explicitly, but the calculators it put up online will let you decide for itself. APC has developed a range of sophisticated guides and tools for IT managers trying to benchmark the efficiency with which their data centers use the appalling volume of electricity they burn through every year.
The rest of the industry is pitching in on the let’s-save-money effort as well. Red Hat is expanding open-source v-tools, Cisco is pitching in with beefier virtual-network management, Citrix is recasting its interpretation of application virtualization, and various experts are offering their own expertise and advice on how to manage the whole virtual world.
All this is kind of catch-up. Tools and guides on virtualization have been a bit thin on the ground so far.
Neither the experts or vendors have gotten much into the blowback, and the potential negative impact of a wide-ranging, money-saving, efficiency promoting technology.
They haven’t often mentioned that, once you finish your initial virtualization project, you’ve still got a lot of work in front of you, not only in adding the kind of management and other refinements that show up in Phase 2 of any major project, but also in adapting your IT staff and the rest of your IT resources to the virtualized infrastructure.
That’s going to be really, really hard if the CEO and CFO see virtualization primarily as a way to save money, cut staff and limit IT projects that involve either capital expenses or new infrastructure.
(“What do you mean we still need to upgrade the network backbone? Didn’t you just cut out half the servers that were plugged into it? Why do we need to buy new servers, can’t you just use the extras left over from the consolidation?”)
We’re collecting blowback stories and tips from early virtualization adopters who have gotten over the organization rough spots as well as the technical. Share your experience as well. What has your experience been following virtual-server tests or large-scale installs? How much blowback or new demands have you faced?
Let us know in the comments below or write me directly here.
In the meantime, avoid discussions about politics, sex or religion at company gatherings, and if pressed for a serious explanation of a fault that’s not your fault, well, just blame Canada. As a scapegoat it might not make much practical sense, but it’s direct, simple and easy to dance to.