Confluent Cloud
Connect a Confluent Cloud organization once, import its Kafka clusters with their Schema Registry, ksqlDB cluster and managed connectors, and manage them next to your self-managed clusters.
Before you start
- A Cloud API key of the organization, created with
confluent api-key create --resource cloud. KCM uses it to discover resources and manage connectors; it cannot read or write topic data. - A Kafka API key scoped to each Kafka cluster you import.
- Optional: a Schema Registry API key and a ksqlDB API key to import them too.
- For metrics, the MetricsViewer role on the Cloud API key.
- A KCM administrator account.
Import clusters
- Open Settings > Confluent Cloud, or Import from Confluent Cloud on the clusters page, then Add organization.
- Enter a name, the Cloud API key and its secret, and click Test. The test reports the access to the Metrics, IAM and Flink compute pool APIs. The secret is stored encrypted and never shown again.
- Explore the organization and pick an environment. KCM lists its Kafka clusters, Schema Registry and ksqlDB clusters, and shows which ones are already imported.
- Import a Kafka cluster: choose the KCM environment and the cluster name, and enter the Kafka API key.
- Optionally import the Schema Registry and a ksqlDB cluster with their own keys, and enable the managed connectors. KCM checks every key before saving.
Partition reassignment, leader rebalancing, preferred leader election, what-if simulation, broker configuration edits, SCRAM users and client quotas of the Kafka protocol are hidden, because Confluent Cloud manages them. Topics only offer the settings Confluent Cloud lets you edit, and the replication factor is set by Confluent Cloud. ACLs are managed for service accounts, users and identity pools, picked by name.
Managed connectors
Once enabled, Confluent Cloud connectors appear in Kafka Connect with a Confluent Cloud badge. You can check their status and task errors; create and edit them (secrets masked by Confluent Cloud must be entered again, and KCM never stores them); pause, resume, restart and delete them; and read, alter or reset their offsets. Each Confluent Cloud cluster with managed connectors enabled uses one Kafka Connect slot of your license.
Metrics and service limits
Topic and partition sizes, produced and consumed throughput, and cluster health tiles (partitions against the limit, active connections, retained bytes, cluster load on Dedicated clusters) come from the Confluent Cloud Metrics API. Data is delayed by about 3 minutes and covers the last 7 days. The applied service limits and their usage are shown on the dashboard and in Settings.
IAM, API keys, role bindings and quotas
Enterprise On a cluster linked to its organization, the Security page adds tabs next to the ACLs, all through the Cloud API key of the organization:
- API keys of the cluster: create a Kafka API key for a service account (its secret is shown once and never stored by KCM), search, delete (except the key KCM itself uses).
- Role bindings (Standard clusters and above): DeveloperRead, DeveloperWrite, DeveloperManage and ResourceOwner on topics, groups and transactional ids (literal or prefix), Operator and MetricsViewer on the cluster.
- Client quotas (Enterprise, Dedicated and Freight clusters): produce and consume limits per service account or identity pool, plus one default quota.
- Access panel per principal: its organization, environment and cluster roles and API keys, where cluster role bindings can be added and removed.
- RBAC analysis: what each role allows, redundant role bindings that are safe to remove (and the broader one to keep), least-privilege hints.
- Settings > Confluent Cloud: service accounts, users and identity pools of the organization.
Roles of the organization key: OrganizationAdmin covers everything. Otherwise grant AccountAdmin (service accounts, users), ResourceKeyAdmin on the clusters (API keys of other accounts) and CloudClusterAdmin (role bindings, quotas). In KCM, reading needs VIEWER on the cluster, changes need cluster ADMIN, and organization-level screens are for KCM administrators.
Flink SQL
Enterprise Compute pools, statements and a SQL workbench for the clusters of the organization. See ksqlDB and Flink SQL.