Kubernetes as the great equalizer has transformed the way we define application infrastructure. One of the many Kubernetes benefits is that you can simply describe/declare what you need in a YAML-based configuration and Kubernetes is off to the races to fulfill your request.
Though the ease and ability to tune with a single line of YAML the resources you need, sometimes we are prone to over-provisioning for safety. Container-based workloads can get expensive which we go into detail about exactly how Kubernetes becomes expensive due to rocketing consumption.
The main goal of Cloud Cost Management is to democratize cloud cost and the mystery around usage for all. As an engineer, how many times have you seen metrics on your running containerized workloads outside of your local or development environments? The irony of that statement is usually you only see items when you exceed a threshold, not if you are well below a threshold. The manifestation of underutilization is a higher cost which is directly what Cloud Cost Management is geared to help with. Now, Cloud Cost Management can recommend the best ways to take action.
Recommendations Are Here
We are very excited to announce a big enhancement to our Cloud Cost Management product, Recommendations. Cloud Cost Management can now give you recommendations on the tuning of container resources. Now the guesswork has been taken out of doing math on resource constraints vs ROI; Cloud Cost Management actually shows you the potential cost savings.
Without Cloud Cost Management recommendations, usually, data around tunings can be lost in an organizational shuffle. Feedback loops need to be there from platform engineering/operations teams alerting for underutilized deployments. Then historical data needs to be provided to development teams to make a judgment on potential reductions in resources. If you are reading those statements and wondering how you would aggregate and disseminate that information, you are certainly not alone.
Digging into the recommendations there is two quality of service [QoS] recommendations; guaranteed and burstable.
Both QoS recommendations are computed based on the historical CPU and Memory utilization data. The implementation uses decaying histograms for CPU samples and Memory peaks. The Guaranteed resources are calculated at the 90th percentile. For Burstable resources, lower and upper bound are based on 50th and 95th percentiles respectively. Doing this by hand would certainly be a task that is not fun and easily accomplished with our Cloud Cost Management platform.
Harness, Your Guardian of Cloud Costs
We are continuing to craft and innovate on new cloud cost management features to help you and your organization save money on cloud spend. Recently we also announced the ability to extract cloud cost information via API from the Harness Platform. Before Cloud Cost Management, managing your cloud spend was certainly a consternation. At Harness, we are on a mission to democratize cloud spend and help disseminate valuable and prudent information as quickly as you consume cloud resources. Stay tuned and make sure to sign up for Cloud Cost Management, today.