News Blog /

Azure Kubernetes Service – Cost Optimisation Techniques for ISVs

by Spanish Point - Feb 15, 2024
Azure Kubernetes Service – Cost Optimisation Techniques for ISVs

In today’s fast-paced digital landscape, Independent Software Vendors (ISVs) are continually seeking ways to deliver innovative solutions while effectively managing costs. As more organisations continue to expand their cloud usage, to optimize return on investment and speed up time to market, it is essential to implement an appropriate cloud computing strategy that ensures cost effectiveness. Azure Kubernetes Service (AKS) emerges as a powerful platform for deploying, managing, and scaling containerised applications, and in this article we’ll explore a variety of useful cost optimisation techniques designed for ISVs using AKS.

Sizing your cluster – Use cluster pre-set configuration

It takes careful consideration to select the appropriate VM SKU, number of nodes, number of availability zones, etc. Cluster pre-set configuration highlights the influence on cost and offers a set of recommendations for various situations. For example, mission critical production environment would require higher spec VM SKU with redundancy across AZs (Azure Availability Zones) and Azure monitor turned on, while for Dev/Test cluster things can run light with unnecessary features turned off.

Sizing you workloads – Set resources requests and limits

To manage the compute resources in an AKS cluster, use pod requests and limits. Pod requests and limits inform the Kubernetes scheduler which compute resources to assign to a pod. As it might be challenging to set the appropriate amount for limits and requests, use Vertical Pod Autoscaler to automatically set resource requests and limits on containers per workload, based on past usage. By doing this, it guarantees that pods are scheduled on nodes with the necessary CPU and memory capacity.

Azure cost optimisations

Start Optimising with AKS today!

AKS Cluster Start/Stop

For AKS clusters that don’t need to be running all of the time, such as Dev/Test environment when it is not in use, a Production cluster that is performing temporary work like ML training, by using cluster start/stop you can completely turn off your cluster. By doing this, you can reduce the computation cost while preserving your objects and cluster state for when you restart the system by shutting down all of the node pools (both system and user).

Running Dev/Test workloads or POC with AKS

AKS comes with a free tier that do not charge for control plane, meaning that a fully managed Kubernetes control plane will be provided to you at no expense. This is great if you would want to do a quick POC on AKS, without having to pay for the control plane.

Explore using Azure Spot Virtual Machines

With Azure Spot virtual machine, you may use unused capacity at significant savings on Azure (up to 90% as compared to pay-as-you-go prices). It is a appropriate only for workloads which can handle interruptions such as batch processing jobs, Dev/Test environments, etc.

Optimise node resource usage in AKS

Elasticity, or the capacity to grow or shrink computational resources (nodes) in response to demand, is one of the main advantages of cloud computing. Using AKS correctly can guarantee that there is always sufficient capacity to satisfy demand without going overboard and wasting unnecessary idle resources.

To conclude, these cost optimisation techniques within AKS are not only about saving money, but also unlocking the full potential of your ISVs capabilities. By implementing the strategies outlined in this article, you’re not just trimming expenses; you’re paving the way for sustainable growth, enhanced performance, and greater agility in the competitive marketplace, and with the help of our experts at Spanish Point we can guide your organisation in achieving cost optimisation with AKS.