We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Scaling Kubernetes-based Event-driven Workloads with Keda & Karpenter • Roland Barcia • GOTO 2023
Scaling event-driven workloads with Keda and Karpenter: Learn how to optimize performance, cost, and flexibility using these open-source projects for Kubernetes-based applications.
- Keda (Kubernetes Event-driven Autoscaling) is an open-source project that helps scale Kubernetes-based event-driven workloads.
- Carpenter is an open-source project that helps scale Kubernetes nodes based on demand and cost optimization.
- Keda acts like a metric server, exposing a rich set of metrics and events to various scalers.
- Scalers can be connected to different provisioners, such as AWS, GKE, or AKS, allowing for customization and flexibility.
- Keda provides a way to define a scaled object, which is a configuration that specifies how to scale an application based on metrics.
- Carpenter provides a way to define a provisioner, which is a configuration that specifies how to provision nodes based on demand.
- Keda and Carpenter can be used together to provide a scalable and cost-optimized solution for event-driven workloads.
- Keda and Carpenter provide multiple options for scaling, including horizontal pod autoscaling, vertical pod autoscaling, and cluster autoscaling.
- Keda and Carpenter can be used with various event sources, such as SQS, Kafka, and Cosmos DB.
- Keda and Carpenter provide a way to define self-managed nodes, which can be used for cost optimization and flexibility.
- Keda and Carpenter provide a way to define node groups, which can be used for scaling and cost optimization.
- Keda and Carpenter provide a way to define node labeling and tainting, which can be used for scheduling and resource allocation.
- Keda and Carpenter provide a way to define node groups with different machine types, which can be used for scaling and cost optimization.
- Keda and Carpenter provide a way to define node groups with different provisioners, which can be used for customization and flexibility.