Common Concepts of Impala Application Development

Users can access the server from the client through Java APIs and Thrift APIs to perform Impala-related operations. The Impala client in this document refers to the ImpalaImpala client installation directory, which contains sample codes for Impala access using Java APIs.

Hive Query Language (SQL-like statement, similar to Hive)

The StateStore checks on the health of all Impalad instance in an Impala cluster, and continuously relays its findings to each of those Impalad instances. If an Impalad instance goes offline due to node failure, network error, or other reasons, the StateStore informs all the other Impalad instances so that future queries can avoid making requests to the unreachable Impalad instances.

The Catalog Service relays the metadata changes from each Impalad instance to other Impalad instances in a cluster. The catalog service avoids the need to issue the REFRESH statement on other instances when the metadata changes on an Impalad instance. When you create or modify a table in Hive, you need to issue REFRESH or INVALIDATE METADATA.