MapReduce Service

MapReduce Service

Learn More

MapReduce Service

One lake per enterprise: large, fast, converged, and reliable
Large: Up to 20,000 nodes per cluster.
Fast: All data kept within the lake, faster data retrieval and analysis.
Converged: Full convergence of batch, streaming, and interactive data analysis, unified resource scheduling, over 90% resource utilization.
Reliable: Hitless upgrade for the latest technology at zero downtime.

Real-time, incremental data updates, offline and real-time data warehouses over the same architecture
• Data imported into the data in real time: from T+1 to T+0.
• One copy of data for multi-purpose data analysis.
• The offline data warehouse can be seamlessly upgraded to a real-time one, allowing for converged batch and stream analysis.
• Decoupled storage and compute, EC ratio as low as 1.2, a TCO reduction of over 20%.

Integrated data lake and warehouse
• The in-lake interactive engine outperforms same-class products by over 30%. You can generate BI reports by yourself using the data in the lake through a self-service interface.
• Convergence of batch, streaming, and interactive data analysis via a unified SQL interface.
• Collaborative computing across MRS and GaussDB(DWS), no need to move data around.