Last time we talked about how, even with popular IaaS pricing readily available in on-line calculators, you don’t really know what a public IaaS service is going to cost until you’ve tried it out, or at least carefully modeled it. Now let’s say you’ve done that. You know how many and what type of instances you need, how much and what type of storage, how much bandwidth, and you have a good estimate for things like disk I/O that you will also be charged for. Using the published rates you can estimate what the service should cost per month. But before you decide, “Provider X gives me the lowest number, so that’s the best choice financially,” a few other things to consider include:
The Future. You may have a good idea what the service will cost in the first month, but what about in the 20th month? If the needs of the application change substantially then your monthly payments will change as well. Will Provider X still be the lowest-cost option? Some IaaS providers drop their prices frequently as their own unit costs decrease. Do you have any assurance that they will continue doing that? If you reserve infrastructure for a longer term to get a better price, will you benefit from those price drops?
The Scope. If Provider X has a lower price, but you have to do twice as much management with your own people than you would with Provider Y, then the less expensive choice may actually be Provider Y. Make sure your NPV analysis includes all the costs and not just the provider’s charges.
The Control. IaaS providers typically want to bill you for usage of the service, similar to, say, how phone service is sometimes billed. For long distance you might pay by the minute for the minutes you actually use. Ever hear a horror story about a family teenager that talked all week and ran up an astronomical bill? That was because the family couldn’t manage their demand. The same goes for cloud – if you can’t manage your demand then you can’t predict your invoices, and you may need to avoid usage-based pricing. This is the kind of thing that will drive your financial people crazy.
The Elasticity. What happens when you release infrastructure, say at the end of a project or a seasonal low point in your business? Does the amount you pay scale down just as rapidly as your consumption? Are there limits on how far you can go? You might be surprised by how many cloud services do place limits on this.
The Flexibility. The way you are charged for services today may not be the way you would like to be charged tomorrow. Do you have alternatives in case the option chosen turns out not to be right for you? What about the invoices themselves – can they be broken down in a way that facilitates your internal chargeback? Can you easily verify them against the resources actually consumed?
All of these can have a significant impact on which service is actually the optimal choice for your needs, so make sure to keep them in mind when you look at pricing.