Never buy Platinum or Silver support again

I can't take credit for the following insight; it comes from one of our Account executives who speaks with IT executives every day.

First, an analogy: When you get a flat tire you don't call a mechanic and expect them to send a truck to diagnose the tire, patch/repair it on the roadside, and re-install it. Instead you carry a replacement spare, and worry later about what to do with the flat one.

But in contrast, IT pays scads of $$ each year for the equivalent of the truck to rush on-site and fix the tire (usually termed sub-12-hour "platinum" support) . If a system goes down, IT ops expects the serviceperson to show up on their doorstep and diagnose & fix on-the-spot. The cost to do this this is absurd -- sometimes as high as 40% of the original system cost -- annually.

So, instead what we see is an end to paying for "platinum" support. It’s a scenario where IT carries "universal spares" -- essentially bare-metal servers that can be re-purposed and re-provisioned in a matter of minutes. BTW, this does not imply that they need to be virtual, since we have special IP that can re-provision physical-to-physical instances of software. Set the failed hardware in a maintenance pool, and just have the serviceperson swap-out the hardware once/month. That should add-up to some savings.

More Data Center Energy Efficiency Surveys

Back in January, I posted a blog with data from a survey Cassatt conducted with hundreds of IT professionals about energy efficiency. I also posted a follow-up blog citing a number of other surveys.

So, in keeping with the theme, I've found a few more resonating surveys and pieces of data.

First: Ken Brill of The Uptime Institute just published a survey on "Data Center Capacity and Energy Efficiency" indicating some amazing factoids about "the invisible crisis" that's looming:

• 42% claimed their data centers would run out of power in 12-24 months
• 23% claimed their data centers would run out of power in 24-60 months
• 39% claimed their data centers would run out of cooling in 12-14 months
• 21% claimed their data centers would run out of cooling in 24-60 months

Next: Paul McGuckin of Gartner Research just published the 2007 Data Center Conferenct Poll Results for Power and Cooling Issues. The first of many findings tables tells quite a story:

Other high-points from Gartner:

• Power capacity has overtaken cooling capacity as the #1 facility problem in their current data center. Gartner attributes this to the high-density equipment being purchased and rolled out.
• Companies must do some sort of facility-related action to deal with this problem (including expansion, upgrade, and even relocation). Only 10% said they would take no action.

Net-net: It would appear that many data centers have ~ 2 years or less of runway left. Given that most new data center projects take a year or more to fund and build, it would seem that 2008/9 are the years that the industry will really begin to make some changes.

Comprehensive Slide Decks on Energy-Efficient Computing

This very cool e-book by Matt Stansberry on TechTarget's SearchDataCenter covers green computing and data center energy efficiency -- and explains the forces driving IT energy consumption, why you should care, and how you can mitigate the problem.

In particular the 5th installment really sums things up:

Chapter 5: A step-by-step guide to greening the data center

The previous four chapters of this data center efficiency eBook laid out the case for going green in the data center and the tactics to achieve energy efficiency across the server, storage and infrastructure environments. This final chapter helps IT managers pull it all together and execute on a data center efficiency project, step-by-step.

• Step 1: Convince the C-Suite to pursue energy efficiency. How much do you spend on energy now? How soon will you have to build a new data center to keep up with your company's growth? Use that information to get a mandate to pursue efficiency from the C-Level executives.
  
• Step 2: Measure data center energy consumption. Set up a simple metric like Power Usage Effectiveness to get a baseline of how much power is going to the servers and how much is being lost on cooling and infrastructure. Set goals to improve the ratio.
  
• Step 3: Get started with IT asset management. Tackle IT inefficiencies by consolidating applications, deleting deadweight servers, implementing active power saving features and specifying energy efficient hardware.
  
• Step 4: Tackle facility fundamentals. Implement hot-aisle/cold-aisle, seal up that floor, raise the voltage on your PDUs from 120 to 208 and take advantage of free cooling when available.

There are also some very clear operational recommendations regarding power management, as well as "active power myths" that dispel the traditional objections to certain operational conducts.

Postcards from IT Financial Management Association Meeting

I am writing this on my way back from a speaking engagement at the IT Financial Management Association’s meeting in Austin TX where I discussed some new metrics and perspectives for measuring costs in shared IT infrastructures from Cassatt's perspective.

The ITFMA is an undiscovered gem-of-a-conference *if* you’re on the finance side of the house supporting IT. Heavy on the analysis, discussion and interaction, and light on the vendor pitches (unless you’re IBM). With between 200-300 attendees, this could be one of the last ‘hard-core’ conferences for IT (financial) professionals. In my talk, only a handful of folks in the audience described themselves as technical – all others were in the controllers/CFO office. And this was substantiated by looking at the attendees list.

The 4 day conference was broken down into four tracks: IT financial management for controllers/CFOs, IT asset management, IT performance management and benchmarking, and Government IT financial management.

For the first time it was clear to me how technology was only part of the challenge for enterprises – managing the business, cost-accounting, asset-management, compliance, etc. is just as much of a burden. To that end, there were a number of recurring themes that kept re-surfacing, even for the one day I was present:

• ITIL and ITSM (IT Service Management): process is a big deal, especially to the bean counters. While most think ITIL handles technical processes like config. Management, it’s just as applicable to service and financial management.
• CBDBs – especially as they related to asset tracking and working capital
• Chargebacks – lots of conversation about these, but I bet only ½ to 1/3 of attendee companies actually do this
• True costs of IT – a major recurring theme. How cost is computed, allocated, and controlled.
• Change management – many agreed that change management is the largest cost sink-hole around. Controlling and tracking changes – especially unplanned changes – is the largest headache IT financial managers face

To the last point, a book called “The Visible OPs” (available via the IT Process Institute) was heralded as a great way to look at IT financial operations through the eyes of ITIL.

From a speaker perspective, there was a really healthy ratio of "end user" war stories: Steve Rossi, the CFO from the state of North Carolina; Arlan Holmes, the deputy CIO from the state of Missouri; Daniel Elvera, the director of IT Finance from First Data; Patricia Harceg, AVP from Wachovia; and dozens more.

The next ITFMA meeting is June 2-6 in Riverside CA - with another down-n-dirty agenda including more on ITIL and chargebacks.

More validation of server power management

The Green Data Center Blog just found a report-out by the Microsoft Research community on "Auto-Shift", essentially energy-aware server provisioning. Hmm... this sounds familiar.

To me, the report further validates the "server power management" space I've been speaking about -- i.e. turning-off temporarily idled servers, and re-starting them when needed (or per policy, schedule, etc.).

Most sources agree that an idled server chews-up between 60%-80% of its fully-loaded power; a massive waste of energy & cooling during off-peak hours. And, most authorities (DOE, EPA, Uptime Institute) also agree that intelligent power management has to come to the data center.

Stay tuned for some numbers & statistics proving-out the actual performance and savings of this technology.

Open-Source Music?

You'll recall that Radiohead broke with the music industry a few months ago by offering its latest album for free (or for whatever visitors want to pay) online, directly from its website. Then, today I read in in The Register that the Nine Inch Nails are taking an alternate model: offering physical albums at prices that range from free to $300 depending on the package. They are giving away the first nine cuts, as 320 kbps MP3 files, along with a 40-page PDF book that covers the entire album. For $5, fans can get the remaining 27 songs and have the option of getting the files in lossless formats. More money will get you things like albums signed by the band.

Clearly the music industry is experimenting with alternate business models.

So here's the musing I've had for some time -- what about "open sourcing" music? Here's the concept: The artist records the content and distributes it however they want. But then they also issue the raw, unmixed tracks. (fee? free?) They allow anyone else to take the raw "code" and re-mix it however they like. Maybe they'll even be improvements (!) on the original.

There are some interesting semi-precedents to this idea. Back in 2003 the Rolling Stones issued a record of 7 remixes of Sympathy for the Devil (granted it was sold as a pre-remixed, pre-recorded album, with remixes done by a number of well-known DJ's). Then there are those numerous mashups - usually unofficial, un-endorsed re-combinations of two (usually very different) songs... and sometimes the mashup is better than either of the two originals. (My favorite is Nirvana vs. Michael Jackson, called "Smells like Billy Jean")

Anyway, think about this model: artists could complement their original work, and maybe even build an increased following, by "opening-up" their original recording to allow other creative individuals to modify or enhance the original music.

Let's see who the first group to do this will be.

Postcards from IDC's Green IT Forum

I just got back from the 1-day Green IT Forum IDC held in NY last week. I have to say it was pretty-well attended by the IT director/VP types from NY financial firms and the like.

Matt Eastwood and Vernon Turner opened the day with a refreshingly broad definition of "Green IT" -- rather than harping on the usual myopic chats about power and cooling. Rather, their definition includes Energy, materials, recycling, etc., and the intersection of Corporate Social Responsibility, economics and technology. They have a very comprehensive framework to encompass a holistic approach to an organization's "greenness". It involves 9 areas:

• Power & cooling
• Flexible employment
• Economic development
• Reverse logistics
• Diversity
• Philanthropy
• Supply-chain ethics
• Community/NGO involvement
• Carbon trading

What's great is that they are not confining the conversation to the data center perspective; rather, they are looking at how IT technology can impact work - telecommuting, telepresence, etc., as well as many of the traditional IT efficiency approaches. (BTW, in a former life, I helped Sun Microsystems launch their OpenWork practice of flexible/remote work - which is credited with saving the company scads of energy, office space, and employee time).

Steve Sams (IBM's VP of global technology services & facility services) also gave an insightful talk of how they've been approaching the efficiency/economics issue over the years. Part of the problem, he observed, was that 77% of Data Center ops people *don't* pay their own electric bills... so where was their economic incentive to improve? This of course implied that the financial relationships of IT need to change to properly incent/reward the people consuming the energy and generating the waste.

Most sobering -- to me anyway -- was the talk given by Vijay Sankaran (director of infrastructure operations) at Ford. This guy had the 'open kimono' delivery, admitting to the fact that they (like all large enterprises) have thousands of Intel, AIX, HP, Sun etc. etc. servers around the globe, and that he's facing a multi-stage multi-year effort to trim and simplify. Admittedly, he said, there is no point-solution that will help; rather it was a matter of standardizing on fewer platforms, consolidating data centers, and implementing strong process control. And only then will they achieve the types of efficiencies where they can call themselves "green".

I have to say that I was somewhat unimpressed by the VMware break-out presentation, which sounded like every other VMware presentation I've attended. (But it was amusing to hear an audience member ask about pricing pressures caused by a number of other virtualization entrants -- and to hear the VMware rep essentially capitulate to the fact that prices for VMs will come down.) And, although I had high expectations for a presentation from Wachovia, I was also pretty unimpressed by their talk as well.

Closing the day, however, there was a reasonably interesting panel made up of representatives from Credit Suisse, Wm Wrigley Jr. Co., Citigroup and Pepco Holdings. These folks reinforced Matt & Vernon's opening points -- each of them was taking a significantly holistic approach to environmentalism and corporate social responsibility, and IT efficiency was only a part of the imperative.