Applying friction theory to the realm of data storage, Jon Toigo finds public storage clouds to be inevitable due to their reduced transaction costs and management issues.
One of my guilty pleasures during the summer is squeezing in a little light reading that's completely off topic. This past summer, I became enamored with the writings of Oliver Williamson, who was awarded the 2009 Nobel Prize in Economic Sciences.
I know, it's not exactly 50 Shades of Grey or the summer blockbuster reads of past years, but I find digging into the work of Nobel laureates mind-expanding. In most cases, what I'm reading drives me on a quest for more understanding, setting up research into other published works.
OK, so I was reading what Williamson had to say about "friction" and it dawned on me that I was reading about storage technology and the infrastruggle. Williamson and his peers are looking into friction not as a physical force resisting motion, but as a function of financial markets, specifically pertaining to transaction costs within a market. Before your eyes roll, or you jump to another article or column in this fine publication, let me oversimplify the economists' view of friction.
All transactions in a market are impacted by external and internal realities (costs) that create friction. Examples of external transaction costs are pretty obvious. You want to buy a product like a storage rig. You need to consider asset depreciation and tax liabilities that accrue to the transaction: the government wants its piece. You need to deal with shipping details, such as how soon the rig will arrive. You need to consider delays due to component supply-chain issues. Then there's the cost of value-add software and the need to ensure that the functions you're buying work with the type of controller specified in the purchase order. The list goes on and on, but you get the idea. External transaction costs exert a force on the purchase, resulting in a kind of friction -- less the heat and wear-producing kind, unless you're thinking about it metaphorically in terms of headaches and stress.
That leaves internal transaction costs, the other sort of friction-producing phenomena that Williamson and company are tackling in their research. Internal transaction costs are pretty broadly described, but usually distill to the impact of a decision, or set of decisions, to buy and deploy proprietary gear that doesn't consider fitness to purpose or common management -- in other words, the way most of us have been cajoled by the industry to buy our storage kit today.
Internal transaction costs accelerate as storage is operated. Year two: the bill comes due for all that value-add software you bought on your rig. (Year one was free as a deal sweetener.) Then there's the cost of adding disk. It turns out that you can only buy the disk and trays from the vendor. A SAS drive is just a SAS drive, but the vendor needs to sign it, dramatically increasing its cost. After year three, your warranty and maintenance agreement runs out and you discover that renewing the agreement will cost as much as buying a new rig, plus your current rig is already "end of life" according to the vendor's practice of EOL-ing gear 17 months after release.
More important than the internal transaction costs associated with gear operations, you may experience an increase in friction as a function of lack of management. This is especially true if you have an enterprise infrastructure with kits sourced from many vendors (or in some cases from the same vendor). Now you need more people to manage the gear, and it doesn't help that your favorite server hypervisor vendor wants to add more storage segregation and management nuances through its preferred hypervisor management tools.
In the world of economics, this kind of friction produces one of two outcomes: When internal transaction costs exceed perceived external transaction costs, businesses tend to contract. In terms of services, like IT, they outsource more, generally because someone is telling them that doing so will reduce friction. From a storage standpoint, that makes public storage clouds sound pretty inevitable to a lot of analysts -- we've been building storage without concern for fitness to purpose or management for a long time, after all.
Conversely, when external transaction costs exceed the perceived cost of internal transactions, businesses tend to grow -- to take more externally performed work onboard, where it can be conducted with less friction. With respect to external storage or service provisioning (today's clouds, yesterday's ASPs and SSPs, or the service-bureau computing plays of the 1980s), the model seems to sour soon after the real friction becomes evident.
A final insight I found telling in my light reading around economics this summer: Williamson said that friction is usually accompanied by "limited rationality and opportunistic behavior," which often determine "behavior and decision-making." Maybe that's why Frost & Sullivan is running ads in Forbes extolling a 54% compound growth in cloud service provider revenues by 2015. If true, it's the silly season.
About the author:
Jon William Toigo is a 30-year IT veteran, CEO and managing principal of Toigo Partners International, and chairman of the Data Management Institute.