Apache Impala

Apache Impala is a massively-parallel SQL engine for data in HDFS, S3 and object storage, sharing its metadata with Hive through the Hive Metastore. It speaks the HiveServer2 Thrift protocol, so AddisDB connects with the same driver it uses for Hive — on Impala’s own HTTP port.

Who it is for

Impala exists to make queries over a Hadoop-era data lake interactive. Where Hive was built for throughput on long batch jobs, Impala keeps daemons resident and targets seconds rather than minutes over the same tables and the same metastore.

If you already run Hive, Impala is usually pointed at the identical data. Choosing between them is a question of latency and concurrency, not of where the data lives.

Set up the coordinator

1. Note a coordinator host — any Impala daemon acting as a coordinator accepts queries.
2. Use the HiveServer2 HTTP port, 28000. Impala’s binary Thrift port is 21050 and the web UI is 25000; neither is what AddisDB connects to.
3. Check that HTTP transport is enabled on the coordinator (--hs2_http_port).
4. Note the database, and the credentials if the cluster uses LDAP authentication.

# The HiveServer2 HTTP endpoint, not the binary Thrift port
impala-shell --protocol=hs2-http -i coordinator.example.com:28000

Connect from AddisDB

1. New Connection → Apache Impala. Port prefills to 28000.
2. Enter the coordinator host and the database.
3. Add a username and password if the cluster authenticates over LDAP.
4. Set SSL mode to require if the coordinator terminates TLS. Test, then Save.

What AddisDB gives you

Queries and the Big Data Console, the schema tree, exports and saved queries — everything the Hive connection offers, since it is the same driver underneath.

Row limits are applied while results are fetched rather than by wrapping your query, and the driver pages through result batches until it has enough, so a limit is respected without depending on the server’s own fetch size.