Simulate Financial Investment
There is a good chance that each data center growth option has a different proportion of CapEx and OpEx. It can be nearly impossible to accurately model the 10 year outlay for each option, let alone accurately compare between different options. We recommend creating cost estimates for each option including requesting bids for each and every contract before detailed design has even begun.
Using a Bayesian algorithm to account for the uncertainty over time it is possible to compare the 10 year outlay of each option. The predictions will mimic a ten day forecast in the weather report with about the same amount of accuracy (for better or for worse – we can’t predict tornadoes, but we do a pretty good job with hurricanes). So, over the course of 10 years the predictions grow less and less accurate, which should be accounted for in a realistic way if we’re being honest – and we strive very hard to be honest.
The weather report shows a 40% chance of severe thunderstorms for Sunday. Fortunately, we have technology that allows us to plan ahead – if there’s a wedding scheduled for Sunday then we can make sure to book a tent. A tent is pretty primitive technology (we’re talking cavemen here) but it is technology.
Similarly, if we’ve got good intelligence that a disruptive technology is going to be deployed in three years, it’s worth attempting to factor that into our calculations. Similarly, if your company is on track to suddenly increase the percentage of virtualized servers in your data centers or implement low energy servers it is certainly worth attempting to factor that in. However, Your company may opt out of those improvements for myriad reasons.
But making this simulation is no mean feat. In order to simulate financial investment we must determine the initial cost of each option including migration and duplicated infrastructure. We must determine the long term operational costs including leases, staffing and maintenance. We must also attempt to discern the relative reliability of each option on the table. It is possible to build a data center rated with an Uptime Institute Tier IV certification, but this often isn’t worth it. Likewise, the infrastructure of the COLO sites should be investigated.
Once these data have been gathered and the numbers crunched, we can begin to create a forecast that will compound the uncertainty year over year in the net present value of each option. We can then have a very realistic comparison through which we can be very clear about the future.
Will it be 83 degrees next Sunday? Probably not. Will it be pretty close? Very likely. Since you know that there is only a 10% chance of rain, maybe you won’t have to invest in that tent.