FLEET OPERATIONS

One console.
Up to 20 clusters.

Operate a bounded, server-configured Kafka fleet with isolated clients, caches, rate limits and mutation capabilities for each stable cluster ID.

See fleet operations Configure a fleet

Demo workspace · Synthetic data

4 configured clustersServer-owned inventory

Clusters43 writable · 1 read-only

Brokers12Across configured fleet

Topics128Internal excluded

Known lag13.5kBounded coverage

Cluster	Provider	Brokers	Topics	Partitions	Status
EU Production	Strimzi	3	46	288	Healthy
US Production	Bitnami	3	42	252	Healthy
Staging	Apache Kafka	3	24	144	Healthy
Disaster Recovery	Strimzi	3	16	84	Read-only

Broker storageEU Production · 3 mapped brokers

Observed

Broker 0

61%

Broker 1

54%

Broker 2

58%

Filesystem usage from configured Prometheus source

Consumer lagKnown / sampled assignments

Investigate

fraud-screen-v224 / 24 partitions known

12,480

order-orchestrator12 / 12 partitions known

640

inventory-reserver18 / 18 partitions known

228

Deployment-owned connectivity: the browser selects only a stable configured ID. Bootstrap addresses, credentials, namespaces and Kubernetes endpoints stay on the server. Inventory changes require reviewed configuration and a rollout.

LAG-FIRST OPERATIONS

Prioritize the queues that need attention.

Consumer-group summaries show total lag, worst partition and known/total coverage. Missing or uncommitted offsets remain unknown instead of becoming zero.

Consumer groups

Search and sort states, members, committed offsets, sampled end offsets, total lag and worst-partition lag.

Broker storage

Keep PVC requests, provisioned capacity and observed filesystem capacity separate, with source and freshness visible.

Rebalancing signals

Review leadership and replica distribution, preferred-leader election safety, JMX movement signals and Strimzi proposal state.

BROKER STORAGE

Capacity, pressure and fill rate - with sources attached.

Kubernetes PVC data provides requested/provisioned capacity. Optional Prometheus observations add filesystem used, available and growth. Optional JMX adds traffic, under-replication and offline-replica signals.

When a source is unavailable or coverage is partial, BetterKafka shows unknown, stale or incomplete - not a reassuring zero.

Kubernetes

Pod-to-PVC mapping and declared capacity

Built in

Prometheus

Observed filesystem usage and 15-minute change

Optional

Broker JMX

Ingress, egress, reassignment and replica signals

Optional

CONFIGURED FLEET

Bring the clusters that belong together.

Start with one trial cluster, then price the exact paid capacity your fleet needs.

Start free trial Price your fleet

One console.Up to 20 clusters.