Kafka Resource Analyzer

Get an overview of the chaos in your Kafka cluster — without connecting to it.

Paste the output of your topic and ACL dumps and get an instant analysis: topic hierarchy, partition and replication consistency, config audit, and an interactive access graph of your ACLs.

Everything runs 100% locally in your browser — your data never leaves your machine. No cluster connection, no upload, no installation. The example below is the (entirely fictional) Kafka platform of a wind farm operator — replace it with your own dumps.

Topics empty

Plain name list, kafka-topics.sh --describe output, or Strimzi KafkaTopic YAML — auto-detected.

Apache Kafka bin/kafka-topics.sh --bootstrap-server broker:9092 --describe
Strimzi kubectl get kafkatopics -n <namespace> -o yaml

ACLs empty

kafka-acls.sh --list output or Strimzi KafkaUser YAML — auto-detected.

Apache Kafka bin/kafka-acls.sh --bootstrap-server broker:9092 --list
Strimzi kubectl get kafkausers -n <namespace> -o yaml
100% local — your data never leaves the browser

What it shows you

  • Topic hierarchy — your topic names, split at their separators and rendered as an explorable tree. Spot naming conventions (and violations) at a glance.

  • Consistency — how many topics use how many partitions and which replication factors are in use. Hopefully just one.

  • Config audit — every topic config that is set, with its value distribution. For example: cleanup.policy — 10× compact, 5× delete, 20× not set. Expand any value to see the culprits.

  • ACL analysis — who is allowed to do what: per principal, per resource, operation distribution, and an interactive access graph.

  • Cross-check — load topics and ACLs together to find topics without any ACL and ACL patterns that match nothing.

About the example data

The preloaded example is the completely fictional Kafka platform of a wind farm operator: SCADA telemetry flows in, Kafka Streams applications detect anomalies and forecast power output, and alerting, maintenance, settlement and ERP integrations hang off it. Three generations of teams have left their traces — deliberately: almost every warning the analyzer can raise has a specimen in there, from the test-turbine-data topic with replication factor 2 to the legacy SCADA account with wildcard ACLs.