09:00 AM

Public vs. Private Cloud: Dollars & Sense

The move to the public cloud in financial services has been slow, but economics shows it is about to speed up.

A headline in the 2013 report from Everest Group and Cloud Connect on financial services cloud adoption sums up the state of understanding of the technology economics (and other factors) of public versus private cloud: "The Public Cloud Conundrum — Are Buyers Getting It Wrong?"

A few statements from the report are also worth a mention:

  • "There seems to be an overwhelming preference for private clouds across workloads."
  • "We wonder if buyers are overestimating the challenges in public cloud adoption."
  • "There is a significant case for the industry to educate enterprises around the business case for public cloud."

Going beyond surveys and perception, a technology economics perspective adds a valuable dimension to this discussion. First, it appears that a dominant industry trend is to move from today's state of mainframes and distributed computing to:

  1. Private clouds within the enterprise datacenter, which then moves to
  2. Managed private clouds in the enterprise datacenter, which then moves to
  3. Hosted private clouds in a third-party datacenter, which then moves to
  4. Shared cloud services across enterprises, which then moves to
  5. Public cloud services

If organizations aren't exactly following this pattern, they typically are planning some variation of this roadmap. Using this as a model, we can start to overlay some aspects of technology economics, but first we need definitions for cloud, private cloud, and public cloud.

The National Institute of Standards and Technology's (NIST) cloud definition seems to capture the essence of cloud: "Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapibly provisioned... with minimal management effort or service provider interaction."

The public cloud model collides with 60 years of conventional IT wisdom and behaviors. Perhaps for now, the early adopters will be in a position of competitive advantage.

Add on the notion of "private," and you get an environment that is not subject to large-scale sharing and typically is housed within an enterprise datacenter behind a firewall using today's models (or evolves into steps 2 and 3 above). Add the notion of "public," and you are in the land of Amazon's AWS, IBM's cloud services, or those of other providers. Clearly there are myriad variations of private and public cloud that fall under the umbrella of "hybrid."

You may wonder why this discussion of definitions underlies a discussion of economics. The reason is the pivotal concept of the "shared pool of configurable computing resources." And in the economics of sharing, scale is a major factor, as is elasticity.

A simultaneous consideration of sharing and scale in the context of private and public illuminates the dynamics of the situation.

Scale is a major factor when it comes to cloud. The large cloud providers have scale that can't be matched by a single financial services company, or even a group of financial firms. "Recent research indicates that Amazon's EC2 cloud is made up of almost half a million servers," ZDNet reported in March 2012. Estimates for Google and Microsoft are for more than 1 million each.

Let's contrast this with the scale of the nation's four largest banks, the largest of which has a little more than 100,000 servers in total; the sum of all the servers in these four banks is likely 350,000.

In short, enterprise private cloud environments are likely to be subsets of the total enterprise server environment -- definitely far below the 87,500 average of our big banks.

Public housing
At the public cloud level, scale economics plays a significant if not astounding role, driven by standardization, productivity/personnel costs, datacenter costs, and the cost of the boxes themselves. The underlying economics of servers changes radically as a firm moves from in-house private to public.

Gartner IT Key Metrics (2013) data (registration required) shows an average Windows Server OS instance cost of $5,641 with an OS instance to full-time equivalent ratio of 51.1-to-1, with personnel being 49% of total costs. Bring this up to the scale of the public cloud, and you find that total costs are easily two magnitudes lower, and that support ratios are in the tens of thousands-to-1. In addition, factor in the NIST definition and the "pay by the drink" model, and you have cost elasticity that simply cannot exist for an in-house private cloud. Also, consider other public "clouds," such as the power grid or even retail banking, and their economic and social impact.

At one time, electricity was an innovation. Before power grids, a factory that could avail itself of its own generating facility and install electric-powered equipment could attain competitive advantage through automation and operational efficiency. However, this involves the cost of owning and maintaining a generating plant. Once the "grid" arrived, the economics changed, and this innovation was available to all enterprises at likely a lower cost than the dedicated plant model. At that time, owning a power plant became a competitive disadvantage, rather than an advantage.

In summary, the economics of public cloud are quite appealing. The sense of public cloud should be clear (though concerns of security, privacy, and resiliency need to be factored in). Public cloud offers elasticity and economics unattainable in the world of internal IT. It also offers market-driven innovation and evolution. But the model itself collides with 60 years of conventional IT wisdom and behaviors. Perhaps for now, the early adopters will be in a position of competitive advantage. But soon use of public cloud will be at the core of a firm's technology economics -- and not a differentiator, except for those that hold on to the past. That will not be the kind of differentiation any IT leader would want.

Dr. Howard A. Rubin is a Professor Emeritus of Computer Science at Hunter College of the City University of New York, a MIT CISR Research Affiliate, a Gartner Senior Advisor, and a former Nolan Norton Research Fellow. He is the founder and CEO of Rubin Worldwide. Dr. Rubin is ... View Full Bio
Comment  | 
Print  | 
More Insights
Newest First  |  Oldest First  |  Threaded View
Greg MacSweeney
Greg MacSweeney,
User Rank: Author
6/26/2014 | 8:16:05 AM
Re: The value of scale for public cloud is overstated
great additional analysis. However, I wonder how many IT leaders at financial firms are actually thinking about these things when it comes to their cloud investments? Somehow, I think they are using much simpler metrics.
User Rank: Apprentice
6/25/2014 | 12:24:45 PM
The value of scale for public cloud is overstated
Dr. Rubin,

It seems the basic driver of compute economics has more to do with converting Moore's Law into economic benefit - rather than scale economics.

Ultimately, the challenge for public or private cloud rests with efficiently converting Moore's Law into hard-dollar saving, and the availability of that productivity is easily accessible to small and hyper-scale providers.

Drilling into public cloud economics, AWS derives it financial benefit by high-rates of growth - scale itself is the result, not the driver.

High rates of growth produce a lower average server age of inventory.  In simple terms, AWS's ingestion rate of higher per unit capacity or speed (Moore's Law) allows it to reduce fixed capacity instance prices.  That is, it allows the public cloud provider to slice physical server inventory capacity into more instances.

As overall growth rates slow, aggregate average server age increases and that impacts the ability to reduce prices.      

Scale may indeed offer AWS the opportunity to "bleed" losses longer than smaller cloud operators, but ultimately  20-40% annual productivity gains from Moore's Law trumps lesser margins.    For example, if Boeing produced aircraft with fuel efficiency that improved 40% per year, would United Airlines fleet scale really be advantage as it aged?  Or could more nimble market entrants, with newer aircraft, offer a competitive price?  Who could bleed losses longer? 

Basically, success relies on LEAN inventory management, which makes perfect sense since the opportunity cost for holding older server inventory is extremely high.  In this regard, private cloud operators could accomplish the same.

As for the software stack, AWS does realize scale around proprietary cloud OS-type software vs OTS available to small and large players.   But, savvy private cloud operators can also effectively leverage OTS licenses to achieve similar economics.   That's not to say private cloud or financial service firms are savvy operators -- just to say it can be accomplished. 

Farther up the stack, public clouds offer little competitive advantage for application software, and in fact, these costs represent very significant sunk costs for private cloud operators.   Re-engineering these applications to run in AWS -- a more stateless environment, is no small investment.

More Commentary
Gartner: 75% of Mobile Apps Will Fail Security Tests Through 2015
The rise of BYOD means enterprises must implement security testing and containment solutions, according to new Gartner research.
Chip & Pain, EMV Will Not Solve Payment Card Fraud
Switching to EMV cards will lower retail fraud, but it's not enough. Here's the good, the bad, and the ugly.
With UCITS V, $9T Isnít as Easy as It Used to Be
With UCITS V's restrictive remuneration rules and hidden costs, going global may get a little less attractive.
Banks to Increase IT Spend on Big Data Challenges, Finds Aite Report
Big data has presented the greatest challenges and dissatisfaction for banks, yet it is the most likely to see upward spending in the next two years.
Scotland Independence Vote: Haggis & Fragmentation
Scottish independence has far-reaching consequences for the global financial markets.
Register for Wall Street & Technology Newsletters
White Papers
Current Issue
Wall Street & Technology - July 2014
In addition to regular audits, the SEC will start to scrutinize the cyber-security preparedness of market participants.
7 Unusual Behaviors That Indicate Security Breaches
7 Unusual Behaviors That Indicate Security Breaches
Breaches create outliers. Identifying anomalous activity can help keep firms in compliance and out of the headlines.