Optimizing spend and cloud usage is a top priority in today’s economic climate. As the industry shifts towards Kubernetes, EKS is one of the biggest items on your AWS bill.
Kubernetes configurations are often set once and not reassessed for months or until there is a problem, leading to waste and inefficiency. That’s why many are turning to intelligent scaling solutions such as Karpenter, which offers significant advancements in cluster awareness and customization.
However, Karpenter is often tuned as a one-and-done — whereas your ecosystem and market pricing are constantly changing. Many engineering teams are dedicating hours of time and operational expense to constantly rebalancing between RI, SP and Spot.
That’s where Compute Copilot for EKS comes in, constantly tuning your configuration to ensure you are always on the most cost-optimized and stable option, automatically updating your provisioning on the fly if not. Copilot brings:
- Commitment (RI/SP) and total organization awareness to Karpenter
- Spot market pricing and stability awareness to Karpenter
- Effortless Spot savings — Copilot moves instances on and off of Spot for you while ensuring reliability
Copilot now supports YAML configurations, in the latest update for even more flexibility and ease of use. In response to customer demand, this release offers:
Effortless onboarding. YAML configuration makes it ultra easy to onboard your Karpenter-managed cluster to Copilot via UI or API. Simply import your current configurations to migrate to modern Karpenter and put your node provisioning on autopilot.
Full control. Whether you want to delegate your configurations to nOps or exercise granular control, YAML configuration unblocks all Karpenter-supported configuration options.
Maintain configurations with IaC tools
The full extent of YAML functionality, including validation, is also available directly via the nOps API, which users can leverage for managing Karpenter configurations programmatically.
Using the nOps API is as simple as creating an nOps API key and posting new provisioner configurations or configuration changes with a HTTP request.
This use case may be preferable for users who want to maintain their configurations with IaC tools, although our recommendation is to create the provisioner once and let it be managed by our Co-Pilot engine.
Why are teams using Karpenter with nOps?
Compute Copilot was created to address the most pressing AWS cost optimization pain points engineering teams face today.
My cluster doesn’t know how to behave with commitments.
My cluster doesn’t know how much Spot to consume at any given time.
My cluster has limited ability to pick or reconsider stable Spot.
If these challenges sound familiar to you, nOps can help. By automating these complex and time-consuming processes for you, the platform reduces your EKS bill by 40-70%.
Compute Copilot brings AWS organization-wide awareness to clusters, ensuring you fully utilize your commitments and receive a discount on all of your compute
It can be time-consuming and challenging to continually maintain an optimal balance of SP, RI and Spot across various AWS services — particularly in large accounts.
Copilot solves this problem by continually rightsizing, reconsidering, and re-evaluating your workload placement to RI, SP or Spot across your AWS ecosystem. Through frequent metadata ingestion and predictive ML models, nOps identifies underutilized resources and deploys high-priority provisioners to shift usage uncovered instance types to coverable instance types.
Compute Copilot brings Spot market awareness to clusters
EKS workloads exhibit a high degree of variability, making them a challenge to fully cover with commitments. Spot instances offer a compelling solution for managing dynamic workloads by leveraging available capacity for less cost.
Karpenter’s current handling of Spot interaction are limited to the 2-minute termination notices currently provided by AWS, potentially disrupting slow-starting workloads. On the other hand, Copilot uses ML to predict Spot termination 60 minutes in advance. It selects diverse instances with minimal risk of interruption for Enerprise-Grade reliability SLAs.
The result is that you benefit from Spot savings, with all the reliability of On Demand.
More info on the YAML configuration release
Check out these videos with nOps Engineers Hayk Harutyunyan and Thiago Carvalho for more information on the new YAML support options, what they are, and why you should use them.
About nOps
nOps has been entrusted with over a billion dollars of AWS spend, and the platform was recently ranked #1 in G2’s cloud cost management category. Join our customers saving up to 50% by booking a demo today.