Cluster Sizing

The infrastructure supports predefined cluster sizes optimized for different workloads:

Size Node Count Use Case Compute Types
xsmall~13 nodesDevelopment/TestingE5.Flex, DenseIO.E5 (3 digest nodes)
small~23 nodesSmall ProductionE5.Flex, DenseIO2.16 (6 digest nodes)
medium~52 nodesMedium ProductionE4.Flex, DenseIO2.16 (21 digest nodes)
large~78 nodesLarge ProductionE4.Flex, DenseIO2.24 (42 digest nodes)
xlarge~138 nodesEnterpriseE4.Flex, DenseIO2.24 (78 digest nodes)

Each size configures specialized node pools (infrastructure only - applications deployed separately):

  • System: Node pool for Kubernetes system components (CoreDNS, etc.)

  • Digest: Node pool for LogScale data processing with NVMe storage

  • Ingest: Node pool for LogScale data ingestion

  • UI: Node pool for LogScale web interface

  • Ingress: Node pool for traffic proxy/load balancing

  • Strimzi: Node pool for Kafka brokers and controllers (optional)

Note

These are compute node pools with appropriate labels and taints. The actual application deployments (Falcon LogScale Collector, Kafka, ingress controllers) are installed by module.logscale during deployment.

DR Standby Mode: When dr = "standby", the UI and Ingest node pools are automatically disabled (scaled to zero). Only System, Digest, Strimzi, and Ingress pools remain active. This reduces cost while maintaining the ability to quickly bring the standby cluster online during failover. The Ingress pool stays enabled to support load balancer health checks.