Cloud pricing models can be complex. In fact, it’s often difficult for public cloud users to decipher a) what they’re spending, b) whether they need to be spending that much, and c) how to save on their cloud costs. The good news is that this doesn’t need to be an ongoing battle. Once you get a handle on what you’re spending, you can automate the cost control process to ensure that you only spend what you need to.
By the way, I recently talked about this on The Cloudcast podcast – if you prefer to listen, check out the episode.
All Cloud Pricing Models Require Cost Management
The major cloud service providers – Amazon Web Services, Microsoft Azure, and Google Cloud Platform – offer several pricing models for compute services – by usage, Reserved, and Spot pricing.
The basic model is by usage – typically this has been per-hour, although AWS and Google both recently announced per-second billing (more on this next week.) This requires careful cost management, so users can determine whether they’re paying for resources that are running when they’re not actually needed. This could be paying for non-production instances on nights and weekends when no one is using them, or paying for oversized instances that are not optimally utilized.
Then there are Reserved Instances, which allow you to pre-pay partially or entirely. The billing calculation is done on the back end, so it still requires management effort to ensure that the instances you are running are actually eligible for the Reserved Instances you’ve paid for.
As to whether these are actually a good choice for you, see the following blog post: Are AWS Reserved Instances Better Than On-Demand? It’s about AWS Reserved Instances, although similar principles apply to Azure Reserved Instances.
Spot instances allow you to bid on and use spare compute capacity for a cheap price, but their inherent risk means that you have to build fault-tolerant applications in order to take advantage of this cost-saving option.
However You’re Paying, You Need to Automate
The bottom line is that while visibility into the costs incurred by your cloud pricing model is an important first step, in order to actually reduce and optimize your cloud spend, you need to be able to take automated actions to reduce infrastructure costs.
To this end, our customers told us that they would like the ability to park instances based on utilization data. So, we’re currently developing this capability, which will be released in early December. Following that, we will add the ability for ParkMyCloud to give you right sizing recommendations – so not only will you be able to automatically park your idle instances, you’ll also be able to automatically size instances to correctly fit your workloads so you’re not overpaying.
Though cloud pricing can be complicated, with governance and automated savings measures in place, you can put cost worries to the back of your mind and focus on your primary objectives.