E-MapReduce:Select a product form

Enterprise-grade fully managed data platform: You can start job development without building your own infrastructure. It provides a full range of features, from task development and debugging to O&M.

High-performance compute engine: The built-in Fusion Engine delivers performance up to four times higher than open source Spark. It supports the Remote Shuffle Service Celeborn, which provides petabyte-level shuffle capabilities and reduces computing costs.

High scalability and elasticity: Built on Alibaba Cloud's serverless foundation, it provides automatic scaling of resources in seconds. You are billed for the computing resources you use, which reduces costs.

Resource observability and security: Provides job-level and resource-level monitoring and alerts. It is deployed based on a VPC and supports fine-grained access control.

Open architecture and ecosystem integration: Seamlessly integrates with OSS-HDFS, Data Lake Formation (DLF), and DataWorks. It supports the separation of computing and storage and lets you share centralized metadata.

One-stop development experience: Provides built-in version management and isolation between development and production environments. It supports job development, debugging, publishing, and scheduling.

Running Spark jobs using a fully managed service without having to manage cluster resources.

Scenarios that require high elasticity and on-demand billing to quickly obtain computing resources during sudden job peaks.

Scenarios that require a high-performance Spark and a data lakehouse platform to support data lakes for data lakehouse or AI workloads.

Fully managed and O&M-free: Ready to use out of the box. It provides visual instance management, intelligent monitoring, and health diagnostics to significantly reduce O&M costs.

Enterprise-grade cloud-native kernel (Stella): Based on a decoupled storage and compute architecture, it is deeply optimized for performance and stability in this architecture and for lakehouse analytics. It supports isolation for multiple compute groups to ensure resource isolation for critical services.

One-stop data management platform: Integrates features such as an SQL editor, permission control, materialized views, import tasks, and SQL diagnostics to improve development and management efficiency.

High-speed query performance: The Massively Parallel Processing (MPP) distributed architecture and fully vectorized execution engine improve query performance by 3 to 10 times.

Intelligent query optimization: The cost-based optimizer (CBO) automatically selects the optimal execution plan to improve the efficiency of complex queries.

Real-time data updates: Columnar storage supports data import in seconds, atomicity, consistency, isolation, and durability (ACID) transactions, and efficient upserts to meet real-time analytics needs.

Data lakehouse analytics: Intelligent materialized views are automatically synchronized to simplify the extract, transform, and load (ETL) process. It natively supports efficient queries on data lake formats such as Paimon, Iceberg, Hudi, Delta Lake, and Hive.

Multidimensional OLAP analysis: Supports flexible multidimensional analysis and ad hoc queries. It is widely used in scenarios such as operations reports, user personas, metric platforms, and BI analytics to help you make data-driven decisions.

Real-time data warehouse: Provides data ingestion and updates in seconds. It is suitable for business scenarios that require low latency, such as real-time inventory monitoring, order tracking, user behavior analysis, and real-time risk control.

Lakehouse processing and analytics: Seamlessly connects to data lake storage such as OSS and HDFS. It lets you directly query formats like Paimon and Iceberg, breaking down data silos and enabling efficient, unified analytics.

Rich components: Integrates open source components such as Hadoop, Spark, Hive, Flink, and Presto, along with proprietary components such as OSS-HDFS, JindoCache, and DLF-Auth. It is suitable for various scenarios, including batch processing, stream processing, and data lakes.

Flexible and controllable: You can log on to ECS instances to perform custom operations. It provides capabilities for rapid cluster creation, automatic scaling, configuration management, and visual O&M.

Stable and easy to use: Uses 100% community open source components optimized for the cloud environment to improve performance and avoid version compatibility issues. You can set up a big data computing environment in minutes and adjust the resource scale with a single click.

Cost-effective: It lets you use computing resources on demand and supports subscription or pay-as-you-go billing methods. It provides automatic tiered storage of hot and cold data and a rich set of O&M tools to reduce total costs.

Automatic scaling: You can dynamically adjust the cluster size based on time or workload. It supports multiple types of elastic resources.

Deep integration: Deeply integrates with Alibaba Cloud services such as OSS, DataWorks, and DLF. You can use EMR as the compute engine for jobs in DataWorks.

Enterprise-grade big data platforms that require full control over computing and storage resources.

Customization requirements for deployment environments and configurations, such as installing specific open source components or adjusting kernel parameters.

Scenarios that require a balance between cost and performance, and support for long-running batch jobs or mixed workloads.

Shared Kubernetes cluster resources: You can run jobs such as Spark and Presto on an existing ACK cluster. You can share container resources with online services and share computing resources across zones.

Cost-saving: You do not need to purchase a separate ACK cluster. You can reuse idle resources of the ACK cluster to run EMR jobs with a single click. This is suitable for hybrid deployments of online and offline services.

Simplified O&M: You can manage online applications and big data tasks through a unified O&M system. You can use the capabilities of ACK and Elastic Container Instance (ECI) to obtain elastic computing resources more quickly.

Optimized experience: It supports adjusting the Spark version at the job level to quickly experiment with new features. It supports seamless switching between ECS and ACK resource models.

Deep integration with data lake architecture: Adopts a cloud-native data lake architecture. It uses ACK for computing and OSS for storage to achieve separation of storage and computing. Metadata is managed by DLF.

You already have an ACK cluster and want to run big data tasks and online applications on the same cluster to achieve resource pooling and elastic scheduling.

Scenarios with high demand for containerized management, requiring a combination of Kubernetes elasticity and the cloud-native ecosystem (such as CI/CD, service mesh, and microservices).

Scenarios where you want to quickly build a data lake architecture and use DLF to manage metadata.