David Merrill

OpenStack Economics

Blog Post created by David Merrill on Sep 14, 2016

A few weeks back a colleague asked my opinion about OpenStack, and if it can reduce costs. There are many sources that talk about the technical, operational, development and agility benefits of OpenStack (especially for hybrid clouds), so I don't need to go there. Clearly, OpenStack (and similar offerings from MS) will influence the move to, and  adoption of,  growth to cloud enablement.


But for me, as an economist, I tend to try to follow the money.



The short answer to "will OpenStack reduce my costs" is the typical "it depends" answer. There is no doubt that the bottom line for software purchase and licensing fees will be reduced. But it becomes an issue or question of shifting costs, as well as reducing costs.


In our work with IT economics, HDS has done thousands of economics assessments since 2002. In the last 4-5 years, many of these assessments have been around identifying and reducing the cost of a VM. With a statistically-significant number of these assessment well documented, I can start to make generalizations around what makes-up the common cost of VM total cost ownership. The chart below is a composite view of my findings on typical VM Costs:


As you can see, the license costs (SW maintenance) tends to be around 12% of VM TCO, while labor tends to be closer to 25% of TCO. What I read, and what I see with customers is the adoption of OpenStack, will increase the development staff required for IT to maintain and manage VMs and the workload. So one quick observation is that reducing the cost of the <12% element> will increase the <25% element>. On the surface, that does not make sense.


Labor is one of the most expensive budget line items with IT, and it keeps getting more expensive. For some geographies, the talent pool is becoming a scarce resource as well. Having an increased dependency on labor is counter to many IT cost improvement programs. OpenStack requires techncally-current skills not only to get started, but to maintain over the life of the infrastructure.


IT organizations need to determine their own cross-over point, at which all the benefits of OpenStack can be off-set by a potential increase in the management and development staff. This cross-over point needs to determined with a multi-year perspective. How much money in savings for SW license or purchases can be re-invested in development and staff.


The move to OpenStack does not wipe-away software license costs. I had a colleague price out 3 years of software costs for a 16 node system. VMware vCenter 6 Std for vSphere 6 with software support was at 'price parity' with an OpenStack SVC Red Hat Premium offering. Now this was a simple comparison, but proved to me that a shifting of costs may be more common that a pure cost reduction. Again, your mileage may vary.


If you are tasked with an OpenStack project, and need a supporting business case, be sure to look at all costs, and not focus on the obvious change to software licensing.


As products and solutions are packaged with OpenStack standard offerings, the dependance on labor will diminish over time. Watching to see how MS Azure and AWS evolve with plug-in and other standards will certainly change the macro economic landscape of IT infrastructure over time.