AWS

Overview

KubeSense collects 591 metrics across 44 resource types from AWS, reading them from Amazon CloudWatch, plus Amazon RDS Performance Insights for database internals.

Collection is read-only and agentless: resources are discovered through the provider's own management APIs, and nothing is installed in your AWS account. Every metric below is requested only for the resource types you enable.

Resources collected

30 resource types are available, 29 of them with metrics. Each row is one toggle on the integration's Resource Types selector; see Resource & Metric Collection for connecting the account itself.

Discovered as lists the entity kinds a toggle produces — the kinds you filter on in the Infrastructure explorer. One toggle can yield several, which is why the 44 metric tables below outnumber the toggles here.

Browse discovered resources

Discovered AWS resources appear under Infrastructure > Cloud Infrastructure on the Cloud Explorer tab, grouped under the AWS provider.

The Resources panel on the left lists the entity kinds produced by the toggles you enabled — the Discovered as column above. Select one to list its resources; the count beside the heading reflects the current filters.

The result table shows the fields that identify a resource — name, account, region, cluster, environment, and status — and the Filters panel below narrows the list by any indexed field, with the match count beside each value. Use the search box at the top to filter the selected resource type by name or field.

Entity kinds from an Inventory only toggle, such as the VPC topology kinds, appear here like any other but have no Metrics tab behind them.

Resource details

Select a resource to open its detail panel beside the list. The header identifies the resource — its provider ID, name, creation time, account, region, environment, and current status — and the arrows step to the next or previous resource without closing the panel.

The panel's tabs depend on what KubeSense has for that resource:

The charts on the Metrics tab are the metrics listed for that resource type in Metrics by resource type below, at whichever tier the integration is configured for.

ECS container and task metrics

The EcsContainer metrics below are collected at a finer grain than the rest of ECS, and under different conditions. Two things decide whether you see them.

They require Container Insights with enhanced observability. Plain Container Insights is not enough: without the enhanced opt-in (available from December 2024), AWS publishes no container-level series at all, and KubeSense discovers nothing to collect. Enable it on the ECS cluster, not in KubeSense. Nothing needs configuring on the KubeSense side to find them — the collector lists what CloudWatch actually publishes rather than guessing, so a cluster without enhanced observability costs no API calls.

They come at two grains. CloudWatch publishes the same container metrics against several dimension sets, and KubeSense collects them as follows:

The per-service grain is the default because its series count is fixed by the number of container definitions. The per-task grain is opt-in because its series count follows the number of running tasks, which changes on every deploy: each new task is a new series, and each is billed per metric on every collection cycle. Turn it on for the services where one task diverging from its peers is a real failure mode, not across an estate.

To enable it, add per_task to the ECS metric extras on the integration. Per-cluster collection is capped, and the cap is reported in the controller logs if a cluster exceeds it.

Metrics by resource type

Tiers are set per resource type on the integration: Core is the default, and Expanded is a superset adding deeper per-service detail. A metric that depends on an optional provider feature — Enhanced Monitoring on RDS, Container Insights on ECS and EKS, Atlas Search, the Prometheus plugin on Kong — returns nothing until that feature is enabled, and KubeSense simply records nothing for it.

ALB target group

Resource type LoadBalancerTargetGroup — 10 metrics.

API Gateway HTTP API

Resource type ApiGatewayHttpApi — 6 metrics.

API Gateway REST API

Resource type ApiGatewayApi — 7 metrics.

API Gateway WebSocket API

Resource type ApiGatewayWsApi — 7 metrics.

Application Load Balancer

Resource type LoadBalancer — 21 metrics.

Auto Scaling group

Resource type AutoScalingGroup — 13 metrics.

Classic Load Balancer

Resource type ClassicLoadBalancer — 13 metrics.

CloudFront distribution

Resource type CloudFrontDistribution — 14 metrics.

Data Firehose stream

Resource type FirehoseStream — 20 metrics.

DocumentDB cluster

Resource type DocumentDbCluster — 24 metrics.

DynamoDB table

Resource type DynamoDbTable — 11 metrics.

DynamoDB table (per operation)

Resource type DynamoDbTableOperation — 3 metrics.

EBS volume

Resource type EbsVolume — 11 metrics.

EBS volume (Nitro instances)

Resource type EbsVolumeNitro — 7 metrics.

EC2 instance

Resource type Ec2Instance — 22 metrics.

ECS cluster

Resource type EcsCluster — 14 metrics.

ECS container

Resource type EcsContainer — 10 metrics.

ECS service

Resource type EcsService — 15 metrics.

ECS task definition

Resource type EcsTaskDefinition — 5 metrics.

EKS cluster

Resource type EksCluster — 11 metrics.

ElastiCache cluster

Resource type ElastiCacheCluster — 19 metrics.

EventBridge bus

Resource type EventBridgeBus — 12 metrics.

Kinesis Data Stream

Resource type KinesisStream — 15 metrics.

Lambda function

Resource type LambdaFunction — 18 metrics.

MSK broker

Resource type MskBroker — 21 metrics.

MSK cluster

Resource type MskCluster — 4 metrics.

Neptune cluster

Resource type NeptuneCluster — 15 metrics.

Network Load Balancer

Resource type NetworkLoadBalancer — 11 metrics.

NLB target group

Resource type NetworkLoadBalancerTargetGroup — 2 metrics.

OpenSearch domain

Resource type OpenSearchDomain — 36 metrics.

RDS / Aurora cluster

Resource type RdsCluster — 3 metrics.

RDS instance

Resource type RdsInstance — 52 metrics.

Redshift cluster

Resource type RedshiftCluster — 12 metrics.

Redshift cluster (query latency)

Resource type RedshiftClusterLatency — 2 metrics.

Route 53 health check

Resource type Route53HealthCheck — 6 metrics.

Route 53 hosted zone

Resource type Route53HostedZone — 3 metrics.

S3 bucket

Resource type S3Bucket — 18 metrics.

SageMaker endpoint

Resource type SageMakerEndpoint — 14 metrics.

SNS topic

Resource type SnsTopic — 9 metrics.

SQS queue

Resource type SqsQueue — 9 metrics.

Step Functions state machine

Resource type StepFunctionsStateMachine — 7 metrics.

Transit Gateway

Resource type TransitGateway — 8 metrics.

WAF web ACL

Resource type WafWebAcl — 10 metrics.

RDS Performance Insights

Resource type PerformanceInsights — 41 metrics.

Collected from Performance Insights rather than CloudWatch, and filtered per instance against the metrics that instance actually publishes. The db.* series follow PostgreSQL vocabulary and vary by engine; the os.* series require Enhanced Monitoring to be enabled on the instance. The aggregate is part of the metric name: .avg is the average over the interval and .max its peak.