Building a Scalable Key-Value Database: Project Hydracus
The world of NoSQL and alternative database implementations (i.e. non-relational) is deeply fascinating to me. I can’t help but be swept up in the whirl of planet-scale web development scalability techniques and the evolution of how developers think about building their applications knowing that with success comes the inevitable need to scale to levels almost unimaginable just five or ten years ago.
I’m going to make a prediction: Developers will be expected to understand the fundamentals of how different database systems can be applied within a singular application; their strengths and weaknesses, and when it is appropriate to leverage them.
There’s a lot out there about scaling relational databases. Partitioning your data over application-managed shards is a topic that has seen its fair share of attention. But I think there’s a whole slew over NoSQL databases that have been built and designed by some very smart people that are new enough that their internals aren’t well understood by most developers to the same level as the typical relational database. My hope is that I’m remedying that need with this new article series and, in the process, building a functional distributed key-value database with the community. I’ll be documenting each stage of the process that I go through during my exploration into designing and coding each piece of this database and, with some helpful guidance from the community, evolving and sharing it for all to use – it will of course be open sourced.
You can read the full story here.