OpsCamp Through an Internet-scale Lens
Like Agile Roots in Salt Lake City in June 2009, OpsCamp in Austin last week demonstrated how powerful grass roots conferences can be. We might not have had big names on the roster, but we sure had a productive dialog on the tricky issues lurking in the cusp between software development and IT operations in Cloud environments.
The conference has been amply covered by Michael Cote, John Willis, Mark Hinkle, and Damon Edwards (to name a few). This post restricts itself to commenting on one fundamental aspect of the cloud which IMHO does not get the attention it deserves. It might be implied in various discourses on the subject, but I believe it needs to be called out as a fundamental assumption for just about anything and everything one might consider doing with respect to the cloud. I am referring to economies of scale.
As pointed out in a forthcoming book on Cloud Computing by colleague and friend Annie Shum, the cloud phenomenon is fundamentally driven by substantial economies of scale in very large data centers. The operational costs of running such data centers are close to an order of magnitude lower than these prevailing in small and mid-sized data centers. User benefits are primarily derived from these compelling economies of scale.
I will be asking Annie to write a detailed guest post on the subject for readers of The Agile Executive. Until her post is published here, I would recommend we primarily consider the Cloud as a phenomenon that only becomes meaningful at scale. In particular, Private Clouds are not likely to yield Internet-scale efficiencies. Folks who regard their company’s conventional data center as a private cloud might be missing up on the ‘secret sauce’ of cloud computing.
The various agile system administration schemes discussed at the Austin OpsCamp are essential to attaining the requisite economies of scale in cloud services. Watch out for follow-on OpsCamps in other cities for developments to come in this all important space.
The Agile Executive 
It seems that the economies of scale are relatively understood in terms of capital expense, power consumption, hardware management, etc. There is still much concern over monitoring and managing virtual infrastructure, and many of the conversations I had with ops managers were centered around these unknowns. While the infrastructure costs equation is somewhat easy to calculate, the costs around headcount, tools, and business impact for missed SLAs are still being worked out. As the thought process matures in these later areas, Internet-based cloud adoption seems likely to skyrocket.
Mike Lunt
February 8, 2010 at 12:28 pm
Hamilton gives a factor of 7.1X for the savings on system administration in very large data centers over small/medium data centers.
Israel
israelgat
February 9, 2010 at 8:06 am
I’m not referring to the system administration at the hardware level, where the costs are easy to calculate. The problem is in “service” administration. In this case, monitoring a business service composed of remote/public virtual infrastructure and SaaS-based applications increases the complexity to the Ops manager. As Ops managers become more comfortable with managing at this layer, the adoption will continue to increase, but in the short term, the cost equation will not be an obvious decision for every organization.
Mike Lunt
February 10, 2010 at 12:58 pm
A very good point and a real source of concern, Mike. Both vendors and clients have been doing Business Service Management for quite a few years now. Public infrastructure has been part of so doing at least since the turn of the century. How come we are staring again at a problem of “service” administration?!
Israel
israelgat
February 11, 2010 at 7:41 pm
Hi Israel,
I might be going against conventional cloud wisdom, but the economies of scale argument is somewhat of a red herring. Sure it gets the CFO/CEO to agree to fund exploratory projects in the short term. Save x% on hardware/software is IT Sales 101 and it works.
However, for all but the smallest of companies, the hardware savings (true savings, not marketing hype) just aren’t significant enough to make a meaningful impact on their business. As one CIO put it, “if I told the board that I all I needed to really improve the business was to rack another $5M of servers, I’d have the money tomorrow”.
So how will cloud technologies have a meaningful impact? By reducing friction and improving agility throughout their IT lifecycle. Reducing the barriers and “instantaneously” getting the developers and testers whatever resources they need. Allowing development projects to ramp up with little to no lead time. Being as responsive as possible to business decisions or changing demands of the marketplace. Add up the value of those benefits and it dwarfs any infrastructure savings.
The barriers of traditional IT infrastructure and IT management processes are far more costly than the inefficient use of hardware budgeta. “Private clouds” can deliver this transformative change just the same as public clouds can.
In the end it all comes down to…. agility! 🙂
-Damon
http://dev2ops.org
Damon Edwards
February 8, 2010 at 5:12 pm
Yes, but… The inevitable question is what price tag the market establishes for Agility. In the long run this price will be determined to a great extent by economies of scale. A factor of 7.1X (see my earlier reply to Mike Lunt) in system administration cost is compelling. Furthermore, the price of entry (for building a very large data center) is so steep that only the biggest of the guys will be able to afford it. Hence, most companies are not likely to achieve the required scale in their own data centers.
A very interesting hybrid is Amazon’s VPC – a private cloud beyond the firewall. I plan to have a post on the subject published in the near future.
Best,
Israel
israelgat
February 9, 2010 at 3:50 pm
I’m in total agreement that there is good value to be found seeking economies of scale.
My point was that this cost savings is an evolutionary benefit to a business (not much different than negotiating better terms with facilities, bandwidth, hardware, or managed hosting providers today). It’ll get cloud technologies “in the door”, but it’s hardly revolutionary.
In the absence of gross technology mismanagement, I think you’d be hard pressed to find cases where a web business beat another competing web business because of lower operating costs. You’ll find plenty of examples where one business beats another because they can out iterate and out maneuver the other.
The revolutionary benefits that cloud technologies and/or cloud purchasing model can bring to a company is helping to enable that increase in agility (i.e. enabling your business to be more flexible and react quickly to market conditions).
Of course, you bring up the difficult question… how do you measure that value/opportunity? Or is it just something that you find out too late when a competitor leaves you in the dust?
Damon Edwards
February 9, 2010 at 11:15 pm
Good points, Damon. Here are a few reflections:
I personally believe we are making big strides toward measuring value/opportunity. B2C is ahead of B2B in measuring value, but I am confident many of the B2C techniques will in good time migrate to B2B application. See A Core Formula for Agile B2C Startups for a quick summary of the state of the art.
I also believe technology assimilation acts as a damper with respect to “The revolutionary benefits that cloud technologies and/or cloud purchasing model can bring to a company…” It is not “only” a matter of changes a company needs to adjust to, but a readjustment of the whole socio-economic infrastructure.
I wish we had the opportunity to speak in length and depth at OpsCamp…
Best,
Israel
israelgat
February 11, 2010 at 7:57 pm
Colleague and friend Annie Shum drew my attention to the following quote from Geoffrey Moore’s Dealing with Darwin:
“Over time, however, competitive advantage within categories shifts inexorably toward volume operations architecture.”
Israel
israelgat
February 13, 2010 at 11:20 am
[…] a comment » Economies of Scale have been much discussed in The Agile Executive since the recent OpsCamp in Austin, TX. The significant savings on system […]
Harnessing Economies of Scale in Cloud Computing to Realize a Greener Computing Option « The Agile Executive
February 16, 2010 at 8:09 am