Alibabacloud Emr Cluster Managev0.0.3

Set up a data lake cluster (DATALAKE) for storing and processing large volumes of structured and unstructured data. User specifies region, components like HDFS, YARN, Spark, and Hive, and selects storage-compute separation with OSS-HDFS for cost efficiency.

When running out of resources in a production EMR cluster, add CORE or TASK nodes to increase storage or compute capacity. Supports auto-scaling configuration for dynamic resource adjustment.

Create a NORMAL mode Hadoop cluster for development and testing. Uses PayAsYouGo payment and local MySQL for Hive metadata to minimize cost and complexity.

Deploy an OLAP cluster (e.g., StarRocks) for real-time analytics on streaming data. Requires high-performance compute and in-memory processing, typically with HA mode and Subscription billing.

Check the current state (e.g., Running, Idle) of one or more EMR clusters, including node-level health. Also handles subscription renewal to prevent service interruption.

Clusters are provisioned on a subscription basis (monthly/yearly) for production workloads, providing predictable costs and reserved capacity. Revenue comes from upfront or recurring subscription fees.

Dev/test or elastic workloads are billed by the hour/minute, allowing customers to pay only for what they use. Revenue is variable based on actual consumption.

Automatically add/remove TASK nodes (often Spot instances) based on workload, reducing costs while maintaining performance. Revenue from premium auto-scaling features or compute savings passed to customer.