CMAK vs Redpanda Console
ComparisonsCMAK vs Redpanda Console, compared
Key takeaway
CMAK needs a ZooKeeper ensemble, and Kafka 4.0 has none, so a KRaft cutover ends it on the day rather than degrading it. Redpanda Console gates four capabilities — including Reassign Partitions, the operation CMAK is most praised for — behind an enterprise licence whose price is not published, and it needs one deployment per cluster, where CMAK registers many in one instance. Its message viewer is the strongest of any free tool. Kpow by Factor House is licensed per cluster at a published price.
What is CMAK?
CMAK is Cluster Manager for Apache Kafka, originally Kafka Manager, built at Yahoo and released under Apache 2.0. It is a Scala application on the Play framework, and it needs a direct connection to a ZooKeeper ensemble to function at all. Its scope is Kafka cluster management rather than data-plane work, and there is no commercial distribution, no hosted offering and no paid support tier.
- several registered clusters from a single view
- creating and modifying topics, and managing partitions
- partition reassignment and preferred-replica election
- optional JMX polling at broker and topic level
The credit is real: reassignment and preferred-replica election are what it is most consistently praised for, and its visual taxonomy of brokers, partitions and replicas maps closely enough to Kafka’s own model that the tool still teaches somebody what a cluster is doing. Nothing has shipped since April 2022.
What is Redpanda Console?
Redpanda Console is an open-source web interface for inspecting and managing Kafka-compatible clusters, built on Go and React and distributed as a Docker image and a Helm chart. It began as Kowl, by CloudHut, and Redpanda acquired it in April 2022. Against a Redpanda cluster it reads the Redpanda admin API for extra capability, and it also serves vanilla Apache Kafka, Amazon MSK and Confluent Platform. v3.11.0 shipped in August 2026.
- Deserialisation: Avro, Protobuf, JSON, XML, CBOR, MessagePack and binary hex, with Protobuf working from local descriptor maps and no registry.
- Filtering: JavaScript message filters, and time-travel offset management.
- Observer Mode: browses a topic without joining a consumer group, so inspection triggers no rebalance and moves nobody’s offsets.
What is the official 2026 pricing of CMAK and Redpanda Console?
CMAK’s licence cost is zero and always was. Its whole cost is what it takes to keep an unmaintained Scala application running: build it from source with sbt, or take a community Docker image, except that both community deployment paths are archived and read-only. There is no support tier to buy, and 522 issues stand open with nobody triaging them.
Console’s community build is free under the Business Source License, indefinitely, covering the message viewer, topic and consumer group management, Kafka Connect management and Schema Registry browsing. What it does not cover is the operations and access half: OIDC and OAuth 2.0 single sign-on, authorization, debug bundle generation and Reassign Partitions all require an enterprise licence whose price is not published. That is a Redpanda Enterprise platform licence rather than a Console product, so a team on vanilla Apache Kafka or Amazon MSK is buying from a broker vendor whose broker it does not run. Fifty people across development, staging and production is three Consoles, because there is no broker-tier multi-cluster, plus a purchase order for the login and the reassignment screen.
Where does each one run out?
CMAK’s hard ZooKeeper dependency is now a date rather than a design preference. Kafka 4.0, released 18 March 2025, operates entirely without ZooKeeper, and the KRaft production readiness proposal is where that removal was settled. A KRaft cutover stops CMAK working on the day rather than degrading it, and managed services reached the same place earlier.
- Data plane: no message browsing at all.
- Freshness: reads come from an internal cache rather than live broker APIs.
- Access control: LDAP basic auth and coarse global feature flags, with no SAML, no OIDC, no per-topic granularity and no audit log.
- Longevity: long-running instances hang after 20 to 30 days from thread-pool exhaustion, reported in 2018 and never addressed.
Console has no broker-tier multi-cluster, so development, staging and production is three deployments, three configuration files and three upgrade paths. It was asked for in September 2021 and again in April 2022, and the later request is still open. Reassignment is an AdminClient operation, settled in the replica reassignment API proposal, so an interface either exposes it or it does not, and here it is licensed.
- Monitoring: no built-in broker metrics, no alerting and no historical trends, so Kafka broker monitoring is a separate Prometheus and Grafana stack.
- Write path: no producing a message from the interface, and no request quotas.
- Timeouts: 35 seconds on ListMessages, and 5 and 6 seconds on DescribeConfigs, DescribeLogDirs and Metadata, all hardcoded.
- Degradation: with one broker offline in a multi-node cluster, all consumer group queries fail.
Which should you pick?
Keep CMAK if:
- the cluster is on ZooKeeper and staying there for now
- the daily work is partition reassignment and preferred-replica election
- nobody outside operations logs in
Pick Redpanda Console if:
- the daily work is reading payloads rather than moving partitions
- Protobuf without a registry removes a real step from incident work
- inspection currently disturbs consumer groups
- either the access model can stay open, or the enterprise licence will actually be bought
CMAK’s date belongs to Kafka rather than to the tool, and deferring the decision does not turn it back into a choice. Where Console is chosen, Kafka monitoring has to already live in a Prometheus and Grafana stack, because Console was never going to cover it. The question underneath both is which half the team cannot do without: one of these is a manager that cannot view, and the other is a viewer that manages up to a licence line.
Kpow: manage and view, without a licence line in between
Both of these do half the job and gate the other half somewhere else. CMAK does reassignment and preferred-replica election, the operation it is most consistently praised for, but it has never browsed a message and it cannot reach a KRaft cluster at all. Redpanda Console browses messages in seven formats and reads the Redpanda admin API, but Reassign Partitions — the equivalent management operation — sits behind an enterprise licence whose price is not published. Kpow by Factor House does both from one stateless container, configured entirely through environment variables, with no external database, sidecar or persistent volume, managing up to 12 clusters from a single instance. It is licensed per cluster at a published price, so adding an engineer never moves the number.
A manager that cannot view and a viewer that manages up to a licence line are both half a tool. Kpow publishes the price for the other half.
Sources
- KIP-833: Mark KRaft as Production Ready
- KIP-455: Create an Administrative API for Replica Reassignment