Me And Clouds
Those who know me or who have heard me speak from time to time know that I am no advocate of cloud computing. I'm not just being obstinate or iconoclastic in taking this position. And I am not hiding under a rock or living in a bubble where news of the real cannot reach me.
It's just that cloud woo carries with it a lot of deja vu -- for those of us who have been in the computing world for 30+ years.
Are you old enough to remember Service Bureau Computing? It was the next big thing...back in the 80's. At the time, a recession was on, companies were asking whether they were in the business of IT or the business of business (they chose the latter, obviously), and many saw the idea of outsourcing their IT infrastructure and operations to a third party provider as a key to solving the riddle of how to benefit from data processing without having to hire all of those annoying data processors.
Perhaps that experience predates your entre into your first data center.
What about this one? Do you remember Application Service Providers or Storage Service Providers (ASPs/SSPs)? They appeared in the late 1990s, initially to support dotcom startups who could not afford much with state-of-the-art IT infrastructure, but later to provide these services to businesses everywhere. After all, there was a recession looming and no business guy thought of his company as a technology company, so why stand up an IT operation of his own? Outsourcing to a qualified ASP/SSP was supposed to be the wave of the future. No more IT staff to pay, no more technology to own, buy your application access by the drink from an on-line provider.
And now, for the past few years, we have been been hearing a lot of talk about something marketing folks call "clouds." This coincides with yet another recession, another round of questioning (belly-aching really) by senior management: if we aren't tech companies, why are we spending so much money on tech? Why not just purchase everything from a cloud service provider?
We have more acronyms now, because they make what is essentially marketecture sound like architecture: IaaS (Infrastructure as a Service), PaaS (Platform as a Service), SaaS (Software as a Service or, alternatively, Storage as a Service), DCaaS (Diaper Changing as a Service)... Basically, anything can be followed by an "aaS" (I'm fond of JackaaS) and its a shiny new thing.
So, what's the problem. Sounds like we have a tradition going. A technology meme steeped in history. As cyclical and predictable as Kondratiev Curves in economics, right?
Well, there are some recurring themes:
- IT labor costs money. In fact, as a percentage of overall IT budget, labor costs are expected to consume about 70% by 2013 according to a stat quoted in a vendor slide deck you may see this week.
- Outsourcing generally delivers significant labor cost reduction in its first couple of years.
- In a recession, companies are trying to reduce overhead and optimize revenue -- decoupling IT expense from income seems like a good idea.
- If guys like Nicholas Carr are correct and IT no longer delivers competitive advantage to companies, why spend any money on doing it yourself?
- Math is hard. Let's go to the Mall. Talking Barbie recording, c. 1999
All good reasons to look into outsourcing. I researched this option at some length in a book I wrote in the early 2000s, The Essential Guide to ASPs.
After interviewing just about every company offering ASP/SSP services at the time (most of which have since gone belly-up), I learned several important facts.
- IT labor is expensive, but returns its investment many times over in the form of overall corporate productivity and decision-making support. Actual mileage varies, but companies with bright techies tend to be more successful than those without them.
- After a couple of years, most companies sour on their outsourcing contracts. In the ASP/SSP era, companies that signed 10 year agreements were usually in court or in front of mediators seeking to break the arrangements within two years.
- Decoupling IT expense from income may seem like a good idea in a recession, until you realize that making leaner staff more productive is essentially possible ONLY by increasing reliance on the smooth operation of the machine. An essential question is whether you feel better having mission critical underpinnings for your business under your direct supervision and control or feel comfortable outsourcing them to a third party provider who counts you among its many many service accounts. Cloud vendors make money by sharing the same service across the maximum number of customers, so in their ideal world, your workload is just one in a cast of thousands. That bodes ill from an odds-making perspective: the more disks you have the more drives break every hour, so the more outages you can expect. Even a short term outage can have damning consequences for the firm. David Lithicum estimates that Amazon lost $4 million in sales during a 49 minute outage in its cloud on Jan 31. Network World's Brandon Butler extends out these numbers and finds that Amazon takes a $5 million/hour hit for downtime. So, you may yield a revenue advantage from skinnying down your labor costs and outsourcing your work, but you increase risk at a macro level. (BTW, later in the piece linked above, the author jokes that we have so much angst about cloud outages while ignoring the "commonplace" outages in our own shops. This thread of reasoning appears in a lot of cloud woo and will be dealt with in the next installment.)
- What if Nick Carr is a JackaaS?
- What happened to talking G.I. Joe's commanding line in the same model year as the talking Barbie: "Let's Go, Men!"
Philosophically, I have a problem with outsourcing without thinking. I have a problem with doing anything without thinking. I have a major problem with outsourcing your thinking -- to a vendor, to a marketing meme... Too many folks seem to be going that way when it comes to clouds.
I will get back to this topic in a bit. It is on my mind for two reasons: first, I am playing the cloud contrarian on an Internet radio program today at 4PM Eastern, and second, I am moderating a webcast tomorrow for UBM and InformationWeek in which IBM will be chatting about its SmartCloud initiative.