It is important to understand the underlying costs for technology investments. Stephanie Overby at CIO magazine states, "Business outcomes from technology investments are all that really matter." IT spend should provide the ability for an organization to achieve or exceed its business objectives.
So what is a business outcome? Typical business outcomes include capital hardware and software avoidance, factory or application uptime, time to market, opening new market segments, optimizing existing markets, ect...
With that in mind, I wanted to take a look at the internal cost structure we created for our bronze SLA tier and compare that to a resource we can purchase from Amazon's EC2. Understanding our underlying costs will help us in our decision making process for deciding which infrastructure components we may want to move to the public cloud.
From Amazon EC2, I decided to go with a small windows virtual machine. I selected an On-Demand instance with (no Contract), which is 0.036 an hour. The 1-year cost for the compute resource is $316.32.
I also included 25 GB of magnetic storage, which comes to $1.25 a month. My total Amazon EC2 Service bill for the (US-East) is $27.61 a month.
I am going to add in some additional costs for running the environment; I have included the Windows Standard License, Norton Anti-Virus License, and heavily discounted OPEX per VM charge for support, which can be seen in the below image. I discounted the infrastructure engineering support costs by 70%, the discount is probably high considering that a significant amount of infrastructure engineering goes into supporting the guest operating system and application integration; but one of the benefits of moving to the public cloud is low OPEX costs according to most analyst.
The total annual cost I come up with is $1,032.63 for a small instance.
If we take a look at our internal cloud pricing model, a bronze level virtual machine with 1 vCPU is $964.62. That is $68.01 cheaper than our Amazon EC2 small instance. The virtual instance includes all the standard charges, matching memory and storage expenses, backup expense, and a 25% discount for OPEX for unmanaged services.
Keep in mind, these costs are for virtual server instances that are running 100% of the time, which is a consideration that needs to be determined when deciding to host a compute resource internal or external. Personally, for most organizations I have worked with in the past, they do not shut down their servers when they are not in use. At the end of the day, I do not see IT professionals powering down their print servers when the workforce goes home.
In the Big Switch by Nicholas Carr, he does a comparison between compute resources and the electric company. I think it is a fantastic book, if you haven't had the opportunity to read it, I highly recommend taking the time to do so; it really sheds a lot of perspective on some of the driving concepts around cloud computing. However, in my opinion, servers are not like electricity, because when I leave my living room at night I shut off my television, cable box, and lights so I don't waste electricity in a metered model. It isn't the same with compute resources. The typical organization leaves their servers running 24/7, and most times their business SLAs incorporate server up-time as a service measurement.
I think the public cloud fulfills specific use cases and should not be ignored. But, getting back to the point of the post, if you don't understand the cost of your underlying IT investments you can't make strategic decisions for the best business outcome. IT spend should be the cause, not the effect, of improved business performance.