Using foxx microservices vs pyarango

Question

I want to set up an api that will query arabgodb. I am debating whether to set use foxx microservices or have a backend that will use pyarango to query the db. What are the pros and cons of both options?

Answer 1

The pros and cons are very much up to you and your skillset (Python vs. JavaScript), but it seems like you are really asking about running your API middleware on a dedicated server (Python) or on within the ArangoDB platform (Foxx).

For single-node installations, keeping your queries close to the server (Foxx) will give you very low-latency query performance, but at the cost of sharing CPU and memory with the DB engine.

The dynamic changes dramatically when using a cluster - the Foxx engine must sit on a different server than the DB engine (see the cluster architecture docs). This relieves the CPU/memory sharing issues and also provides a sort of high availability for your API, but also introduces other, more serious, query performance issues that are unique to a cluster.

Bottom line: if you do not fully understand (or are not prepared to deal with) cluster performance, then stick with a single node. And if you are not willing to give up CPU/memory for your Foxx engine, then use a separate middleware server. There are many other details to consider, but (for me) this is the main decision.

Also, please note that there are only two modes: "cluster" and "single", but several ways to deploy the cluster (master/slave, active failover, one shard, multi-shard, etc.).