Brenda Michelson, Principal of Elemental Links, writes "elemental cloud computing" recently tweeted: "100k buys way more public, than private, cloud computing power" which started a short but inspiring conversation on the subject centering around the observation that "cloud is the gift that keeps on giving." That's alluding to the fact that the compute power purchased in "the cloud" is an annual expense, unlike private, cloud computing power which requires renewal at longer intervals, usually in the 3-5 year range.
Still, Brenda is right at least in the short term. $100,000 purchases a lot more compute power in a public cloud computing environment than it will/would/does in a private cloud computing environment. The problem is that $100,000 in a private cloud computing environment is likely to provide more business value than would a comparable investment in a public cloud computing environment. And that's really the metric we should be using instead of CAPEX versus OPEX.
THIS IS NOT ABOUT CAPEX vs OPEX
I have read, and re-read, and read some more on the whole "CAPEX vs OPEX" arguments both for and against public cloud computing. I started off by reading JP Morgenthal's "A Better Metric for Analyzing the Value of the Cloud". But that led me to digging deeper as it became apparent that my decision to focus on computer science in graduate school rather than economics and accounting was getting the way of understanding the real debate regarding CAPEX vs OPEX. Bernard Golden, CEO of consulting firm HyperStratus, nearly convinced me that public cloud computing is the only financially responsible option in "Capex vs Opex: Most Miss Point About Cloud Economics". Stratetect argued basically the opposite in "CAPEX vs OPEX. What is the difference?" and thus sent me into a rousing debate with myself (I'm still unsure who won that one). I finished up with a quick read of "CapEx, OpEx and Cloudy Accounting" over at Data Center Knowledge because, well, I'm masochistic, apparently.
I finally came to a few conclusions:
- I can explain to you why some CAPEX spending can actually lead to a decrease in overall CAPEX spending but not whether it's desirable to do so. For example, an investment in an application delivery controller can result in higher VM density capabilities per physical server, the result of which is less CAPEX required to purchase physical servers to meet annual growth (and a reduction in the associated OPEX, I might add). Whether this reduction in CAPEX is good or bad seems to be somewhat of a subjective opinion and one I am apparently not well-versed enough in accounting to make.
- Investing in public cloud computing increases OPEX (cost of services) while simultaneously decreasing OPEX (offloads maintenance/management of physical servers, decreases costs associated with datacenter space). Whether the increase is good/bad/desirable is also somewhat of a subjective opinion and open to interpretation. I guess the trick is to try to make sure the decrease in OPEX is greater than the increase, thus resulting in an overall decrease. Bernard Golden insists part of the appeal of public cloud computing is in the ability to just "stop using resources" because it isn't a fixed cost. I disagree that it isn't a fixed cost; it's rare you're suddenly going to not need compute resources, but his argument has merit as a reason why increases in OPEX are more easily dismissed as an argument against public cloud computing.
- There's no good answer to the CAPEX vs OPEX debate. CEOs, CIOs, CTOs, and financial folks all seem to disagree when it comes to cloud computing and costs. In fact I'm wondering at this point whether cloud computing and its impact on the balance sheet might force a change in the equations financial analysts use to value companies. If it's not as simple as "high OPEX bad and more CAPEX good" then perhaps the "economics" of cloud computing will have broader implications than just for technology, IT, and organizations. But for IT and technology-focused individuals rather than focusing on whether one approach is desirable over another we should - or at least I should - probably just stick to explaining how a given solution decreases/increases CAPEX/OPEX and leave the analysis to someone better suited to the task.
Ultimately I think there's no good answer that can come - at least authoritatively - from an outside source. The only person suited to answering the question of whether increasing OPEX by leveraging cloud computing or investing in the CAPEX necessary to leverage private cloud computing is the one sitting in the CTO/CIO chair. Then I reached another conclusion: we're ignoring the very basic question of the business-value offered by each approach. Focusing in on CAPEX or OPEX or X-EX is not nearly as germane as asking "What's the business value an organization is getting when they write that $100,000 check?"
A LINE in the CLOUD
I do agree with Brenda that $100,000 will purchase a lot more compute power in a public cloud computing environment, it is important, I think, to consider the business value of those resources may be far less than if that $100,000 were invested in the servers and infrastructure necessary to implement a private cloud computing environment.
This is primarily because of the way in which those resources can be used.
- They are dedicated. I know this sounds contrary to the cloud computing model, but consider for a moment then when you deploy an application in a virtualized container in a cloud computing environment that the resources allocated for that application are dedicated to that application. You can't share those resources with another application. Sure, you can spread that $100,000 around many applications, but then you're limiting the amount of resources that each application can ultimately consume. If you're investing in physical assets (servers) for a private cloud implementation the compute power you've essentially "purchased" can be shared across many applications, even after the application is retired.
- Infrastructure can't be shared. If you're provisioning infrastructure services in a cloud computing environment those resources can't be shared with the applications residing in your local data center. That means they are essentially dedicated infrastructure for the purposes of serving only cloud computing deployed applications in a single location. Contrast this with infrastructure deployed in the data center, whose services can be shared across all applications - legacy, web, and even virtual private cloud hosted applications.
- Refresh rates mean longer relevancy. Technology refresh rates typically range from 3-5 years. If $100,000 will purchase 3-5 years worth of compute resources, then this is a moot point. If that $100,000 will only provision compute resources in a public cloud computing environment for less than three years and you can purchase enough compute resources with that same budget for a private cloud computing environment, you've got some decision making to do.
- Private cloud computing has benefits public implementations do not. One of the tasks that goes into implementing a private cloud computing model is the codification of IT and business processes via automation and orchestration. That process affords a unique opportunity to re-examine the efficiency of those processes and streamline them, improving the efficiency of IT and, one hopes, the business supported by those processes. This is not as intangible as it sounds and can be measured based on standard labor costs and time to manually perform certain tasks as well as any time savings that may directly impact costs to acquire/retain customers.
The business value of the compute power you're purchasing in a public cloud computing environment may be vastly different than the business value acquired for the same budget and directed toward a private cloud computing environment.
An even better option may be a hybrid model that leverages a virtual private cloud computing deployment along with a private cloud computing implementation. By using public cloud computing resources to extend a private cloud computing implementation you can continue to leverage existing - or new - investments for the private cloud implementation, the virtual private cloud implementation, as well as legacy and more traditionally deployed applications. The business value of compute resources is much higher when it's shared in way as to serve more business needs and its relative cost decreases according to how well it can be applied across a broad spectrum of applications.
This is not a new equation for most IT professionals. Justifying both CAPEX and OPEX is something IT has been asked to do for many years; "bring us a business case" is the typical response to any request for funds. Certainly we should not stop providing the base information: this reduces OPEX, this CAPEX can be used for X because this needs to be used by the people who ultimately make the decision regarding which type of spending is best for their organization, but that's not our call. So instead of worrying about which column of the financial ledger it is best for compute and infrastructure resources to come out of we should focus more on the business value we're offering instead.