Pogocache In-Memory Database

Full Throttle

Pogocache [1] processes more than 3 million requests per second, whereas Memcached only achieves around 2.5 million under the same conditions. At the same time, Pogocache exposes the processor to significantly less load and needs just about half as many CPU cycles as Memcached, which – in turn – reduces server utilization, saving energy and ultimately cash, as well.

The other technical features sound enticing, too. From the outside, Pogocache behaves just like Memcached, Valkey, and Redis and supports the PostgreSQL Wire Protocol. As a result, compatible clients can communicate directly with the database without any modifications, making it easy for administrators to replace a competing product with the faster database. Alternatively or in addition, Pogocache communicates with its own REST interface.

Developers can even integrate the database engine into applications. Requests are sent directly to Pogocache without the time-consuming need to pass first through the network stack. The reported ability to enable more than 100 million operations per second is useful, among other things, for microservices, each of which gets its own fast cache.

The man behind Pogocache is developer Josh Baker, who is responsible for a Redis-compatible server framework, among other things, and has been at the helm of Tile38 – an open source in-memory database for geodata – for several years. The first version of Pogocache was released in July and was already considered stable and suitable for production environments, but the lack of long-term experience and a few caveats suggest that you should not rush in blindly with a Pogocache deployment.

Pitfalls and Limitations

The in-memory database is primarily designed as caching software that maintains a search index. Therefore, Pogocache does not care who is currently retrieving or writing

...

comments powered by Disqus