Until recently, cloud usage reports have predominantly only provided visibility into inventory metrics such as the number of instances that run over time. However, these metrics misrepresent actual compute power consumption since different sized machines with various levels of compute power are treated equally.
On top of measuring inventory, one option to deal with this lack of uniformity is to group instances by type in order to understand usage. However, when dealing with different cloud vendors, or even your private cloud in a multi-cloud, hybrid environment, you need a common denominator in order to run accurate usage and cost analyses. That being said, allow me to introduce ‘Core-Hours’.
Similar to how electric companies meter how much power you use regardless of what appliances are running, we wanted to know how much compute power is consumed by instances. Just as electric companies measure appliance consumption in a specific unit of power (watts), we measure the compute power of an instance by the number of virtual CPUs (cores) it utilizes. And just as an appliance’s total consumption is measured by multiplying its power by the amount of time it runs (typically kilowatt-hours, or kWh), we measure an instance’s total consumption by multiplying its power by the amount of time it runs throughout a specific period of time, hence the name core-hours. So, we created a usage report that is purely based on aggregated instance core-hours, thus providing an accurate representation of the actual compute power you’re consuming. You can leverage the core-hours metric in multiple ways. For example, you can detect trends in consumption in a more accurate manner, or, if you multiply your core-hours by the average cost per core, you can gain insights into your cost of consumption (we’ll discuss more about the average cost of cores in part two of this blog).
We’ve also created a report based on GB-hours of RAM if RAM is the determining power factor (e.g., high-performance databases).
Core-hours provide a uniform way of measuring consumption across all resources in a single place, reducing the complexity of dealing with different resources such as instance families, instance sizes, and cloud providers. Additionally, core-hours have a more granular view of compute power consumption that allows you to see “hot spots” of cores that are not being utilized such as an available four-core instance that can replace four hours of a single one-core instance. This hot spot visibility can be set up according to your applications’ requirements.
Conversely, core-hours make it easier to detect spikes in usage and cloud sprawl since you can differentiate between instance size. This is an improvement from past levels of visibility where you could only see the number of instances, meaning if your instances are scaled up in size, you might not be able to realize until you see your cloud bill. In addition, using core-hours allows you to look at your compute power on a sliding scale, from a general overview to a particular project’s or business unit’s consumption.
As part of the evolving technology of the cloud, using core-hours offers a uniform way of accurately measuring compute power. Core-hours tie together the fringes of cloud management that once made compute power challenging to measure. Ultimately, they provide yet another reason to move to the cloud.
Start Optimizing Your Multi-Cloud Deployment
Start your free trial today to easily identify greater cloud efficiency opportunities and view your entire AWS deployment in one place. Gain full transparency, accountability, and control.