Back to Blog

Eric Wright

The OpenStack Skills Challenge: Cost Comes in Many Forms

There was a recent article that pitted OpenStack against the incumbent virtualization providers on their respective clouds. It was interesting because it seemed to me to highlight a lot of the information that we already know, while also leaving out much of what is also widely known. Data used to feed the article appears to be the idea that a cost per virtual instance can be worked out, and that OpenStack engineers will work out to be more on a cost per-instance basis than other types of virtualization admins. Let’s take a deeper look and see just how we can work with this and come up with a solution, shall we?

What does a Private Cloud Really Cost?

The source article that I’m using is here so that you can review first, which will help. A key quote we can work off of to begin with is this:

"it has been found that VMware, Microsoft and Red Hat all offer private cloud prices within half a cent of each other at $0.10 per virtual-machine hour, whereas OpenStack costs $0.08 per VM hour." (source)

These values have been derived from a number of sources that are mentioned, but only the high level information is gleaned for the purpose of the article. It looks as though things are going well when we see that OpenStack is able to produce a 0.02$ per VM hour reduction in cost. That being said, this is not the only metric that we can use.

Another interesting stat that is used comes from this:

"OpenStack can provide a TCO advantage over the typical DIY approach for deployment. However, this is only where its use results in a 45% manpower saving" (source)

openstackThis raises a very interesting question around where we really need to measure the effectiveness of a private cloud implementation. Just like in manufacturing, we have to be careful how we measure productivity and value. I am always extremely wary of using person-hours in the measurement because there is an implication that we will somehow reduce person counts which is not the goal. The goal of a private cloud is to enable consumers of the service to be more self-sufficient, and to reduce the time to market for application environments. That comes in many forms.

OpenStack as a private cloud platform can appear to be more costly as it is a net-new platform for many organizations. When we look at VMware, Microsoft and Red Hat as the comparative private cloud builds, we will often find a fairly strong base of adoption already in place. Although OpenStack is now 5 years since its launch, this is still relatively young considering the velocity that has been seen in the recent 12-18 months.

My thought on this is that it may be too early to pick a horse for the finish, but we are seeing that OpenStack is among the leaders as it heads into the final stretch of the race, even with its relative youth among the incumbent players.

Who are the OpenStack Users we are Discussing?

reduce-licensing-costsIt is always very important to understand who the target is when a survey is used, or we use phrases like "people we spoke to" and "many are saying" because this can be misleading. Many content sources speak to a target audience who may have a directed view based on some other criteria. I'm careful to note that this article does not present specific detail on people used to feed the content. I don't doubt that there is validity to the numbers being presented, but I am also careful to use this as a true broad, general view of the real cost of the private cloud on OpenStack as this comes in many flavors and styles.

The closing shows an interesting conclusion that is all too often used:

"Comparing the low price of a public cloud to the average cloud, makes this ratio for value even higher, with an engineer needing to support at least 250 VM's." (source)

Remember though, that public cloud and private cloud are apples to oranges. While the concept of consuming the resources on private and public clouds is effectively the same, the scale of growth is usually the driver for public, while security and data gravity is the driver for private cloud.

Much more will be happening as we head into the OpenStack Summit in Vancouver, May 18-22. With the recent Kilo release on April 30th, and much more focus as companies like Citrix jump back into supporting and growing OpenStack, we will inevitably see some different and more attractive price detail on what the true cost of private cloud is. One thing is certain and that is that private cloud is definitely a viable and popularly growing strategy for enterprises. Exciting times ahead indeed!