使用 Homebrew 安装 hive

项目链接

• 主页
• 仓库
• 上游文档
• 软件包管理器页面

命令和别名

• beeline
• hive
• hive-config.sh
• hiveserver2
• hplsql
• init-hive-dfs.sh
• metatool
• replstats.sh
• schematool

项目历史

Facebook engineers started building Hive after data growth exposed the limits of a commercial RDBMS-backed warehouse. The Meta engineering history says Facebook's data grew from a 15 TB dataset in 2007 to more than 2 PB by the time of the 2009 article, and that MapReduce was too low-level for many analysis tasks.

Hive's design brought tables, columns, partitions, and a SQL subset to Hadoop while preserving Hadoop's extensibility. Facebook open sourced Hive in August 2008, and Apache Foundation milestones record Hive entering the Apache Incubator in 2008. ASF milestones record Hive becoming a top-level Apache project in 2010.

The Apache Hive site describes the project as a distributed, fault-tolerant data warehouse at massive scale. It emphasizes SQL over distributed storage, the Hive Metastore as a central metadata repository, HiveServer2 for multi-client access, cost-based optimization, compaction, replication, security integrations, and support for modern storage systems and table formats.

采用历史

Early adoption was intense inside Facebook. The Meta engineering post says Hive was popular with internal users from the start, regularly ran thousands of jobs, served hundreds of users, stored more than 2 PB of uncompressed data, and loaded 15 TB daily.

Open-source adoption followed because Hive lowered the barrier to Hadoop analytics. It let analysts use a SQL-like language while Hadoop handled distributed execution. The Apache site later presents Hive as used by enterprises and cloud/data-platform vendors, and highlights integrations with S3, Azure Data Lake, Google Cloud Storage, Spark, Presto, Impala, Apache Ranger, Apache Atlas, Apache Iceberg, and other data-stack components.

Hive also created a durable ecosystem surface: the metastore became a central catalog for other engines and data lake architectures, while Beeline and HiveServer2 became common access points for JDBC, ODBC, BI tools, and scripts.

使用方式

In package-manager terms, users install Hive to get command-line and service entry points. The Homebrew package exposes `beeline`, `hive`, `hive-config.sh`, `hiveserver2`, `hplsql`, `init-hive-dfs.sh`, `metatool`, `replstats.sh`, and `schematool`.

Historically, users ran HiveQL through the `hive` CLI; Beeline and HiveServer2 became the preferred client/server model for multi-client and authenticated access. Operators configure XML files such as `hive-site.xml`, metastore and server configuration files, and Beeline connection files.

Typical workloads include SQL analytics on distributed storage, ETL, table and partition management, metastore-backed data lake catalogs, batch reporting, compaction, replication, and integration with BI and JDBC/ODBC clients.

为什么软件包爱好者会关心

Hive is one of the packages that made the Hadoop ecosystem approachable to SQL users. It matters in package history because it bridged a low-level distributed-computing substrate and the familiar data-warehouse interface that enterprises already knew how to staff, script, and operate.

The package also illustrates why some CLI packages are really ecosystems. The `hive` formula is not only a command; it packages services, schema tools, metastore administration, connection clients, and configuration conventions. That makes it closer to a platform component than a simple executable.

Hive's long tail is especially visible in the metastore. Even as newer engines evolved, the Hive Metastore remained a shared metadata layer in data-lake deployments, so package maintainers and operators continued to care about Hive compatibility beyond the original MapReduce execution model.

时间线

• 2007: Facebook begins moving large-scale data analysis pressure toward Hadoop-backed infrastructure.
• 2008-08: Hive is open sourced by Facebook.
• 2008: ASF milestones record Hive entering the Apache Incubator.
• 2009-06-10: Facebook publishes the Hive petabyte-scale Hadoop data warehouse article.
• 2010: ASF milestones record Hive becoming a top-level Apache project.
• 2010s: HiveServer2, Beeline, cost-based optimization, metastore use, and enterprise security integrations become part of common Hive deployments.
• 2020s: Apache Hive site highlights data lake, cloud storage, Iceberg, compaction, replication, security, and metastore-centered use cases.

Related projects

• Apache Hadoop is the distributed storage and processing ecosystem Hive was built on.
• Apache HCatalog graduated in 2013 to become part of Apache Hive, adding table and storage management services for Hadoop data.
• Apache Spark, Presto, Impala, and other engines integrate with Hive concepts or the Hive Metastore in data lake environments.
• Apache Ranger and Apache Atlas are documented by the Hive site as security, authorization, lineage, and governance integrations.
• Apache Iceberg is highlighted by the Hive site as a modern table-format integration.

来源

• Apache Hive version-control page for official repository URL.
• Apache Hive website for project description, features, integrations, and adoption framing.
• Apache Software Foundation milestones for Incubator and top-level-project dates.
• Meta Engineering article for Facebook origin, open-sourcing, internal scale, and early architecture.