Redis has revved its JSON support claiming it gives developers a high-speed alternative to disk-based document stores, as well as traditional legacy relational databases.
The real-time, in-memory database vendor added JSON support almost three years ago as one of its first Redis Modules, which extend the core open-source Redis platform.
However, Redis cofounder and CTO Yiftach Shoolman said this week that this initial implementation was very basic: “It allows you to keep a JSON document without being able to do real processing with it.”
JSON 2.0 has been rewritten in Rust, which the company says means higher performance, better stability, and lower memory footprint, and includes support for JSONPath Syntax and JSON Data Interchange Format.
Shoolman said, 2.0 “practically allows you to do everything you want with JSON or document databases, and what’s unique about this is the real-time speed.”
Benefits include atomic operations on JSON sub-elements in memory, and high-speed ingestion of JSON documents, while the native indexing allows devs to rapidly create, populate, and expire indexes for frequently updated JSON data, and perform complex aggregations using groupby/reduce, map sort, and filter.
Redis is claiming some substantial advantages over competing document stores, including 50 times more throughput than MongoDB and seven times more than ElasticSearch on YCSB ecommerce workloads.
Shoolman said there were three key ways to deploy the new module. It could be used for query acceleration on relevant data, or devs could deploy Redis as a front-end database.
Or, he said, users could use the paid enterprise or cloud versions of Redis, which means “you can enjoy the other benefits that we give to our customers, such as using active-active technology, to deploy it across regions and clouds and hybrid or using Redis on flash to deploy large data sets over SSD and DRAM and save on costs.”
Ultimately, Shoolman said, “what we’re trying to do with Redis is to extend the capabilities so developers can start developing any real-time, any modern application from scratch.”
He said that “over 60 percent of the users when they are looking at modern data models – putting aside the core capabilities of Redis – they decide to work with JSON, and this is why we decided to invest in this area.”
