Cloudera CDP Administrator - Public Cloud flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

CDP control plane vs data plane

The control plane (Cloudera-managed) provisions and manages clusters via API; the data plane (customer VPC/VNet) runs the actual workloads and stores data.

Data Lake cluster role

The singleton SDX cluster hosting Hive Metastore, Ranger, Atlas, and Knox; all Data Hubs attach to it for shared security and catalog services.

Data Hub template

A reusable cluster definition specifying services, instance groups, and node types; used to launch workload clusters that reference a Data Lake.

Cross-account IAM role

On AWS, the role in the customer account that the CDP control plane assumes to provision resources; its trust policy must allow the Cloudera account.

CDP CLI access key

A per-user credential (access key ID plus private key) generated in the management console, used by the cdp CLI to call the control plane API.

Recipe

A script executed on hosts of a specific instance group during defined phases (e.g., pre-service-start) to customize cluster nodes.

Environment registration

The step that links a cloud network, storage location, and credentials to CDP; it validates network egress and the cross-account trust before clusters can be created.

Deletion order

Data Hub clusters must be deleted before the Data Lake, and the Data Lake before the environment; the Data Lake cannot be removed while clusters remain attached.

IDBroker

A Data Lake service that assumes cloud roles and issues short-lived credentials to Hadoop users, so workloads never store long-lived cloud keys.

Ranger policy evaluation

Ranger plugins on each cluster poll the Data Lake for policies and enforce them locally; deny policies take precedence over allow policies.

Knox Gateway

The single ingress (CDP Proxy) that fronts service UIs and REST APIs, terminates TLS, and authenticates users before proxying requests.

SDX

Shared Data Experience: the set of Data Lake services (HMS, Ranger, Atlas, Knox) providing centralized security, governance, and metadata across clusters.

Atlas lineage

Metadata service capturing data lineage from Hive and Spark hooks, showing how tables and columns are derived across jobs.

Data Lake HA

A Data Lake deployed with multiple master nodes so HMS, Ranger, and Knox survive a node failure; a non-HA Data Lake is an SDX single point of failure.

SAML/OIDC IDP

An external identity provider used to authenticate CDP users; local user management is the alternative when no external IDP is configured.

Ranger audit store

The destination for Ranger audit events (Solr or an external database), queried for access auditing and compliance reporting.

RAZ

Ranger Authorization for cloud storage: path-based access control enforced on S3/ADLS/GCS paths used by Hive and Impala, brokered through IDBroker credentials.

Centralized Hive Metastore

The HMS runs on the Data Lake and is shared by all Data Hubs, so tables created in one cluster are visible in others.

Impala coordinator/executor

Impala splits into coordinator nodes that plan queries and executor nodes that run them; both read data from cloud object storage.

Instance group scaling

Changing the node count of a specific instance group to add or remove capacity; autoscaling applies load-based policies per group.

Runtime version ordering

The Data Lake runtime version must be greater than or equal to attached Data Hubs; upgrades proceed Data Lake first, then Data Hubs.

Iceberg tables

Open table format using a catalog (HMS or REST) for metadata; queryable by Impala and Spark on CDP clusters.

Livy

A REST service for submitting Spark jobs to a cluster; commonly used for interactive and batch Spark workloads.

Object storage decoupling

Hive and Impala data resides in cloud object storage rather than local HDFS; local HDFS is used mainly for scratch and temporary data.

Cloudera Observability

A telemetry service collecting metrics and logs from CDP clusters for dashboards, alerting, and workload analysis.

Diagnostic bundle

A collected archive of logs and configuration from a cluster or environment, used to troubleshoot provisioning and service failures.

IDBroker 403 error

Cloud storage access denied in Hive or Impala usually indicates an IDBroker failure or a broken trust between the cluster role and IDBroker role.

Failed recipe symptom

A recipe failure during cluster creation blocks service start; inspect the recipe output and the failing host's role logs.

Ranger plugin connectivity

If policies stop applying across Data Hubs, verify each Ranger plugin can reach the Data Lake and that the audit store is writable.

Environment registration failure

Typically caused by blocked network egress or an incorrect cross-account trust policy; verify security groups, NACLs, and the role trust.

Upgrade validation

Run pre-upgrade checks, upgrade the Data Lake first, then Data Hubs; rollback options are limited, so validate in a non-production environment.

Idle cluster cost control

Stopping idle compute Data Hubs saves cost while the Data Lake stays running to keep SDX services available for remaining workloads.