// the find
ClickHouse/clickhouse-java
ClickHouse Java Clients & JDBC Driver
The official ClickHouse Java client and JDBC driver, maintained by the ClickHouse team. It's HTTP-only (no native protocol support), which is intentional — ClickHouse recommends HTTP for most workloads. Two parallel generations exist: v1 (being deprecated) and v2 (the current recommended path), with a separate jdbc-v2 in progress.
LZ4 compression on both request and response out of the box, which matters significantly for ClickHouse throughput on large result sets. Java object SerDe in v2 means you can map query results directly to POJOs without manual column extraction. Active maintenance with JMH benchmarks in CI — the compare-jmh.py script in .github/scripts is a real performance regression gate, not theater. The R2DBC driver exists for reactive stacks, which is genuinely rare among database-specific clients.
v2 drops failover and load-balancing support that v1 had — if you're running ClickHouse in a cluster and migrating to v2, you lose those features with no clear replacement timeline stated. The artifact classifier mess in 0.9.0 (shaded, http, shaded-all all being retired simultaneously) is going to break builds for anyone who pinned those. No native protocol support at all — fine for most uses but rules out some low-latency scenarios where TCP beats HTTP. The two-generation split means the docs and examples are frequently ambiguous about which client they apply to, and you have to mentally track which version you're reading about.