Alibaba Cloud MaxCompute vs Apache Hive

Alibaba Cloud MaxCompute is a web-accessible, cloud-deployed data platform with separate storage and compute. It supports data lakes, streaming ingestion, and in-database machine learning. Standard computing resources cost $22 a month on subscription, with a minimum purchase of 50 CUs. Multi-zone HA computing resources cost $28.60 a month on subscription and provide high-availability compute across multiple availability zones. For elastic capacity, Elastically reserved CUs are listed at $0.05 a month, billed hourly, with scheduled scaling and pay-as-you-go billing. Auto-elastic CUs are listed at $0.06 a month, billed hourly, with usage monitored per second, automatic scaling, and charges based on resources consumed. The pricing model is paid, with a starting pricing note of $0.05. These options separate fixed subscription capacity from elastic resource usage, while the platform’s storage and compute separation supports data workloads that need independently provisioned resources.

Apache Hive is a distributed data warehouse system for reading, writing, and managing large datasets with SQL. Its Metastore keeps shared metadata for tables and partitions and serves clients including Hive, Impala, and Spark; HiveServer2 supports concurrent clients through JDBC and ODBC. Hive provides full ACID transactions for ORC tables and insert-only transactions for other formats. Documented storage options include S3, Azure Data Lake, and Google Cloud Storage, and Hive supports Apache Iceberg tables through a StorageHandler. Named integrations include Spark, Presto, and Impala, with Apache Ranger for authorization and Apache Atlas for lineage and governance. Security options described by the project include Kerberos, fine-grained access control, and audit logging; HiveServer2 also lists LDAP, PAM, and custom authentication. The quickstart includes Docker instructions for the server and Metastore. Hive is free open-source software. The installation manual lists Java 8, Maven 3.6.3, Protobuf 2.5, Hadoop 3.3.6, and Tez among current prerequisites; Hive 4.2.x instead requires JDK 21 minimum and works with Hadoop 3.4.1 and Tez 0.10.5.