RedisJSON Performance Benchmark: Comparing with MongoDB and ElasticSearch

Recent official RedisJSON (RediSearch) performance report demonstrates that RedisJSON dramatically outperforms other NoSQL databases.

Key conclusions: isolated writes – RedisJSON is 5.4× faster than MongoDB and over 200× faster than ElasticSearch; isolated reads – 12.7× faster than MongoDB and over 500× faster than ElasticSearch.

In mixed workloads, real‑time updates do not degrade RedisJSON’s search or read performance, unlike ElasticSearch.

Specific data: RedisJSON supports roughly 50× more operations per second than MongoDB and 7× more than ElasticSearch; latency is about 90× lower than MongoDB and 23.7× lower than ElasticSearch.

RedisJSON maintains stable latency and higher throughput as write ratios increase, while ElasticSearch’s throughput drops.

Query Engine

The development of reresearch and RedisJSON emphasizes performance stability across versions, with tools for analysis and detection.

Version 2.2 improves load and query performance by 1.7× over 2.0, enhancing throughput and data‑load latency.

① Loading Optimization

NYC taxi benchmark results show each new reresearch version delivers substantial performance gains.

② Full‑Text Search Optimization

Indexing 5.9 million Wikipedia abstracts and running full‑text queries demonstrates large improvements from v2.0 to v2.2 in write, read, and search latency, increasing achievable Search and JSON throughput.

Comparison with Other Frameworks

RedisJSON was compared against MongoDB 5.0.3 and ElasticSearch 7.15 using the YCSB benchmark on identical AWS m5d.8xlarge VMs (1 client + 3 database servers) with SSD storage.

MongoDB ran as a three‑member replica set with a text index; ElasticSearch used 15 shards, query cache, and RAID‑0 NVMe SSDs; RedisJSON used RediSearch 2.2 + RedisJSON 2.0 on a 27‑shard Redis Cluster v6.2.6.

Additional network, memory, CPU, and I/O benchmarks were performed to understand underlying VM characteristics.

① 100 % Write Benchmark

RedisJSON’s ingestion speed is 8.8× higher than ElasticSearch and 1.8× higher than MongoDB, while keeping sub‑millisecond latency; 99 % of requests complete under 1.5 ms.

RedisJSON uniquely updates its index on every write, enabling immediate search of newly written documents, unlike ElasticSearch’s near‑real‑time (NRT) indexing.

② 100 % Read Benchmark

RedisJSON reads are 15.8× faster than ElasticSearch and 2.8× faster than MongoDB, maintaining sub‑millisecond latency across the entire latency range.

③ Mixed Read/Write/Search Benchmark

Real‑world mixed workloads (65 % search, 35 % read) show RedisJSON’s throughput equals ElasticSearch’s at 0 % updates, but as update ratio rises, RedisJSON’s throughput remains stable while ElasticSearch degrades.

Latency analysis shows RedisJSON keeps p99 below 0.23 ms for reads, far better than MongoDB (5.01 ms) and ElasticSearch (10.49 ms); for writes, RedisJSON and MongoDB stay sub‑millisecond, whereas ElasticSearch exhibits high tail latency.

④ Complete Latency Analysis

Comprehensive latency percentiles confirm RedisJSON as the only solution maintaining sub‑millisecond latency across all operations, with p99 read latency around 3 ms versus 162 ms for ElasticSearch under 6 K ops/sec sustained load.

How to Get Started

Start using RedisJSON by creating a free Redis Cloud database in any region or by deploying the RedisJSON Docker container; documentation has been updated to help developers quickly use query and search features.

Client drivers for popular languages are available to accelerate adoption.