Table of Contents
1. Total Data vs. Data Egress Costs
2. Total Unattached Cloud Disks
3. Idle VM Instances
4. Provisioned Resources vs. Consumed Resources
5. Untagged Cloud Resources
6. Discounted VM Instances vs. Standard-Price Instances
7. Total Cloud Spending per Day and/or Hour
A Metric-Based Approach to Cloud Cost Control
There's plenty of awareness among businesses today about the importance of tracking cloud spending, especially in economically turbulent times.
But the question is: How, exactly, do you measure the cost efficiency of your cloud? Which cloud finance metrics should you track, and how do you collect the data that drives them?
This article answers those questions by discussing five key cloud spending metrics, including what they measure and how to track them.
1. Total Data vs. Data Egress Costs
Egress fees, which cloud providers charge when data moves out of their clouds, are an easily overlooked cloud cost that, if not properly tracked, can lead to enormous amounts of wasted money.
To manage this risk, calculate your total data egress costs (which you can usually find on your cloud computing bill) relative to the total data you store in the cloud.
Tracking these data points together is valuable because your egress costs will generally increase as the total amounts of data you store in the cloud increase. So, an uptick in egress costs is likely not concerning if your total data storage also grows. But if you see egress expenses going up without a corresponding increase in data, it may be a sign that you're not managing egress efficiently.
2. Total Unattached Cloud Disks
Unattached cloud disks are storage resources that are active but are not being used by a VM. In most cases, unattached disks are unnecessary and should be shut down in order to save money.
By tracking how many unattached disks you have active, you gain insight into how financially efficient your overall cloud environment is.
3. Idle VM Instances
Along similar lines, monitor how many idle VM instances you have running.
Although each cloud provider has a slightly different definition of what counts as an idle instance, they all let you track (usually through their management portals) how many VM instances have been running while using minimal CPU for an extended period. If you have many such VMs, you are probably wasting money on instances that aren't serving any useful purpose.
4. Provisioned Resources vs. Consumed Resources
Another valuable metric for tracking the financial efficiency of your VMs in particular is to compare how many CPU and memory resources they are actually consuming, relative to those you've assigned to them.
In general, your consumed resources should reliably approach (but never equal) total provisioned resources. If you find that total consumed resources are much lower than total provisioned resources — if, for instance, the ratio is below 50% on a regular basis — your cloud VM instances are overprovisioned, and you should rightsize them and/or use autoscaling to achieve better cost performance.
5. Untagged Cloud Resources
Yet another simple but effective cloud finance metric is your total number of untagged cloud resources — meaning resources that have not been labeled with tags that help admins track their purpose.
Untagged resources don't come with any extra fees, so having many untagged resources doesn't necessarily mean you're wasting money. However, because untagged resources are harder to track, this metric gives you a sense of how well-managed your cloud environment is and how efficiently you can find and address resources that are wasting money. In a well-managed, financially efficient cloud, you'll have very few, if any, untagged resources.
6. Discounted VM Instances vs. Standard-Price Instances
You can gain insight into how financially efficient your cloud environment is by tracking how many reserved or spot VM instances (which are available at major cost discounts) you have running and compare that figure to the number of on-demand instances (which don't provide any cost discounts).
Granted, many workloads can't take advantage of discounted resources, so it's not necessarily a bad thing if you're not using them on a frequent basis. But you should at least track how the ratio of discounted to standard-price resources changes over time because if your use of discounted instances trends down, you may no longer be operating your cloud as cost-efficiently as you once were.
7. Total Cloud Spending per Day and/or Hour
Although most cloud providers generate bills on a monthly basis, it's important to monitor your total spending per day to understand how your costs vary from one day to the next — or even hourly, if your workload requirements change very frequently.
In many cases, cloud workload demand will decrease during weekends and off-hours. In a well-managed cloud, spending should also decrease during those periods because workloads should be configured to scale down in order to save money. If your spending metrics remain unchanged even as workload requirements fluctuate, you have room to improve the cost-efficiency of your cloud.
Conclusion: A Metric-Based Approach to Cloud Cost Control
Cloud costs can quickly spiral out of control — but if you're not tracking spending metrics carefully, you won't know you're wasting money, let alone be able to identify the source of your financial inefficiency. Help ensure a healthy cloud computing bill by tracking cloud cost metrics closely and continuously.
About the author
Christopher Tozzi is a technology analyst with subject matter expertise in cloud computing, application development, open source software, virtualization, containers and more. He also lectures at a major university in the Albany, New York, area. His book, “For Fun and Profit: A History of the Free and Open Source Software Revolution,” was published by MIT Press.
