Cloudera CDP Administrator - Public Cloud key terms

A logical container in the CDP control plane that groups a data lake, its Data Hub clusters, and shared services under one set of credentials, region, and network settings. Environments let administrators isolate teams or workloads while reusing common configuration.

The Cloudera-managed SaaS layer that hosts the management console, APIs, and orchestration logic used to create, scale, and delete environments and clusters in your cloud account. It stores metadata but customer data stays in the customer's cloud account.

The foundational cluster in a CDP environment that provides shared services such as SDX governance, storage management, and the metastore that Data Hub clusters attach to. Each environment has exactly one data lake, and it must exist before workloads are deployed.

A workload cluster attached to a data lake, hosting service roles such as Hive, Spark, Impala, or Kafka. Data Hubs inherit governance and storage configuration from the data lake but can be sized and scaled independently per workload.

A reusable definition describing which service roles a Data Hub cluster runs, their instance types, and their scaling policy. Administrators create templates once and use them to deploy consistent clusters across environments.

A named set of cloud instances within a cluster that all run the same roles and share a configuration. Instance groups are the unit for scaling, repair, and node replacement in CDP clusters.

A policy that adds or removes worker instances in an instance group based on metrics such as YARN pending containers or CPU load. It requires a schedule or load-based rule and a defined minimum and maximum node count.

An IAM role in the customer's cloud account that the CDP control plane assumes to provision and manage resources. Its trust policy and permissions determine what the control plane can create, so misconfiguration is a common provisioning failure cause.

A CDP component that issues short-lived cloud credentials to cluster services and users, replacing long-lived access keys stored on nodes. It maps an authenticated identity to a cloud role, enabling least-privilege access to object storage.

The CDP framework that centralizes metadata, governance policies, and security controls across clusters so that definitions made once apply everywhere. It is anchored in the data lake and consumed by attached Data Hub clusters.

The authorization component of SDX that evaluates fine-grained access policies for HDFS, Hive, HBase, Kafka, and other services. Policies are defined centrally and can be scoped by resource path, column, or tag.

A Ranger policy that grants or denies access based on classification tags attached to data rather than on physical paths. Because the policy follows the tag, access stays correct when data is moved or copied.

The metadata and lineage service in SDX that catalogs datasets, tracks how data flows between them, and stores classification tags. Ranger reads Atlas tags to enforce tag-based policies.

The process of assigning sensitivity or business tags, such as PII or confidential, to datasets so that governance policies can target them. In CDP it is often automated by profilers that inspect column contents.

The reverse proxy that exposes cluster web UIs and REST APIs through a single authenticated endpoint. It enforces authentication, applies per-service access control lists, and shields internal hosts from direct client access.

A network authentication protocol in which clients obtain time-limited tickets from a key distribution center instead of sending passwords. CDP clusters use it for strong service-to-service and user authentication.

A configuration in which CDP validates user credentials against an external directory such as Active Directory. It is commonly paired with group lookups so that directory groups map to Ranger roles and policies.

The service that stores table definitions, partitions, and schema metadata for SQL engines in the cluster. In CDP it is hosted in the data lake so that Hive, Impala, and Spark share one catalog.

An open table format that adds snapshots, schema evolution, and hidden partitioning to files stored in object storage. CDP supports Iceberg tables for reliable concurrent reads and writes on data lakes.

A distributed object store that can serve as the storage layer for a CDP data lake, providing an HDFS-compatible namespace at very large scale. It stores data in buckets and volumes rather than traditional HDFS directories.

Cloud-native storage accessed over HTTP APIs, used by CDP for data lake files, logs, and backups. Access is governed by cloud IAM policies and temporary credentials issued through IDBroker.

The interface for defining capacity-scheduler queues, their capacities, and their placement rules. It lets administrators partition cluster resources between teams and is often tuned alongside Data Hub autoscaling.

A mechanism that limits how many Impala queries run concurrently and queues or rejects excess requests. It protects cluster stability by preventing a single user or workload from exhausting memory and CPU.

The CDP operation that replaces or reconfigures an unhealthy instance in a cluster while preserving its role assignment. Administrators trigger it from the console or API after diagnosing hardware or service failures.

The per-cluster management service that deploys, configures, monitors, and restarts the roles running on cluster hosts. Administrators use it for service-level health checks and configuration changes that the control plane does not expose.

The packaged set of open-source components, such as HDFS, YARN, Hive, and Spark, that CDP installs on cluster hosts. Its version determines available features and must match the data lake and Data Hub compatibility matrix.

A collected archive of logs, configuration files, and system metrics from cluster hosts, generated to support troubleshooting. Administrators create bundles before contacting support or after a recurring failure.

A record of access attempts and administrative actions, including Ranger authorization decisions and API calls. Auditors use these logs to investigate incidents and to demonstrate compliance with access policies.

A per-user credential used by CDP clients and APIs to authenticate when Kerberos or SSO is not practical. It is managed in the control plane profile and can be rotated without changing the directory password.

A non-human account created in the CDP control plane for automation such as CI pipelines or scripts. It authenticates with its own credentials and is granted scoped roles rather than sharing a human administrator's identity.

A script or configuration override applied during cluster deployment to customize hosts, install packages, or tune services. Recipes let administrators standardize changes that the console does not natively support.

A setting that routes outbound cluster traffic through an HTTP proxy for environments with restricted internet access. It must allow the control plane and package repositories so provisioning and upgrades succeed.

The controlled process of moving a data lake, and then its attached Data Hubs, to a newer Cloudera Runtime version. It requires compatibility checks, backups, and typically a maintenance window because services restart.

A CDP definition that copies selected HDFS paths or Hive tables from a source cluster to a target on a schedule. It supports disaster recovery and data migration between environments or clouds.

The graceful removal of a node or role from a cluster in which data is drained or replicated elsewhere before shutdown. Doing this before terminating instances prevents data loss and under-replication.

Optional usage and diagnostic data that Cloudera collects from the control plane and clusters to improve support and product quality. Administrators can review and adjust what is shared in account settings.