Answer the question
In order to leave comments, you need to log in
Optimal key-value storage as backend for git objects?
Experimenting on a new project, selecting a database to use as a backend for objects in a git repository (data will be stored as git repositories for history and standard data import/export).
I don’t want to deal with add-ons over the usual git with the file system, I’m thinking about building a suitable infrastructure with libgit2 on top of some kind of key-value storage.
Requirements:
* key-value, binary is stored by hash (string), no complex data types, nested structures, and so on
* reading data by primary key (great if in the form of streams, not all together)
* writing data (data after writing do not change)
* delete data by primary key (very rare)
* easy cluster building (preferably something like specifying a disk where to store, one of the nodes of the cluster - and there, based on the configuration, let it replicate and migrate data as it pleases, plug-and-play as much as possible)
* it is desirable to read and write data, you can it was on any node, let it figure it out on its own as it suits it
* simple API
I didn’t work with such things before, so I can’t make a choice based on experience.
While looking at Voldemort.
Cassandra and Riak seem like overkill for such a task.
Answer the question
In order to leave comments, you need to log in
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question