The Billion Docs JSON Challenge: ClickHouse vs. MongoDB, Elasticsearch, and more
This buries the lede somewhat, but here’s the key bit:
We built a new powerful JSON data type for ClickHouse with true column-oriented storage, support for dynamically changing data structures without type unification and the ability to query individual JSON paths really fast. […] ClickHouse stores the values of each unique JSON path as native columns, allowing high data compression and, as we are demonstrating in this blog, maintaining the same high query performance seen on classic types.
The performance results are very impressive, and notably also efficient in disk space usage.
Tags: clickhouse benchmarks performance json querying columnar-storage mongodb elasticsearch databases storage