Hive新手咨询：Hive Server与Metastore Server的作用及相关疑问

Hey there! Totally get these questions when you're just dipping your toes into Hive—let's break them down clearly:

1. Where is Hive Server, and what does it do? #

When you run the hive command directly and start creating/querying tables, you’re using the Hive CLI (Command Line Interface). By default, this CLI connects directly to Hive’s core components (like the metastore and execution engine) in a local, embedded way—so you don’t need a separate Hive Server running for this basic, single-user use case.

Hive Server (the modern, widely used version is Hive Server2, or HS2) is a standalone service you can manually start with this command:

hive --service hiveserver2

Once up, it typically listens on port 10000. Its main job is to act as a centralized gateway for remote access to Hive. Instead of running the CLI locally, other clients (like Beeline, JDBC/ODBC tools, or even custom applications) can connect to HS2 over the network to execute HiveQL queries. This lets multiple users/apps interact with Hive simultaneously, and you don’t need to install the full Hive stack on every client machine.

2. What's the role of Metastore Server, and can other apps request metadata from it? #

The Metastore is where Hive stores all its critical metadata—think table schemas, database names, partition details, and table location info. By default, when you use the CLI, it connects directly to the Metastore’s underlying database (like MySQL or Derby) in an embedded mode.

The Metastore Server turns this metadata access into a network-accessible service. You start it with:

hive --service metastore

It usually listens on port 9083. Its key roles are:

• Decoupling metadata storage from Hive’s query engine, so you can scale each component independently
• Providing a standardized, secure interface for accessing metadata
• Enforcing access controls to ensure only authorized users/apps can view or modify metadata

And yes—once the Metastore Server is running, other applications can absolutely request metadata from it. Any tool that supports Hive’s Metastore Thrift protocol can connect to it to fetch info like table structures, list databases, or check partition statuses. This is super handy for tools like data catalogs, ETL pipelines, or custom analytics apps that need to interact with Hive’s metadata without running full Hive queries.

内容的提问来源于stack exchange，提问作者mark monkey