This solution adopts EMR to help you deploy clusters with different types of data processing platforms and systems, such as Hadoop and Hive, for data analytics requirements in different gaming scenarios. You can leverage OSS to archive cold data in more cost-efficient storage and store hot data in highly available instances to optimize resource utilization and performance. The OSS + EMR combination architecture can provide speed, reliability, and cost-efficiency consistent with the Hadoop Distributed File System (HDFS). The decoupled storage and computing instances can be scaled separately to adjust system performance and scale flexibly, simplifying management and O&M while lowering upgrading costs.

This solution deploys data processing platforms, such as Hadoop, Hive, and Presto, with highly elastic Kubernetes clusters formed by ECS. You can scale up for traffic peaks during events and promotions or scale down to save costs during low traffic. Query data in various formats are stored in OSS; cold data are archived in OSS Archive, and regularly accessed data are stored in OSS Standard storage for high availability. You can retrieve and manage data with configurable rules, scale the storage capacity for business needs, and optimize storage costs easily. This solution uses Alibaba Cloud Elasticsearch for quick indexing and precise searching of website data and Message Queue for Apache Kafka to monitor website activities and collect real-time statistics. In addition, it uses DLA to process interactive queries and EMR Druid for real-time queries and ad hoc queries. These tasks are processed by the computing clusters (Hadoop, Hive, Presto, etc.), and the results are stored in OSS Standard Storage.