Running JanusGraph on resource-constrained devices

[ncn...@gmail.com]

Hi:

Can JanusGraph run on resource-constrained devices, e.g., Raspberry Pi, or low end PCs with configuration such as 2 GB RAM and 80 GB storage? The primary use cases I'm looking at is the following:
1) Capturing and processing streaming data on the device, converting the data into a graph-based format, and storing it in JanusGraph
2) Periodically moving the stored data onto the cloud via ETL or similar techniques

Thanks.

[Jason Plurad]

I'm not aware of anybody that has tried this. If anybody in the community has, please reply on this thread.

As far as whether it can run, I'd think that it could as long as you are able to run Java 8 on the device. You didn't mention how much data you are planning to store in the graph, but you could possibly get away with going even lighter than JanusGraph by using an in-memory TinkerGraph with periodic writes to disk. If you choose to go down the JanusGraph route, then picking an embedded backend like BerkeleyJE could probably work.

Give it a try and let us know how it works out.

[ncn...@gmail.com]

Thanks. So just to clarify, based on my understanding of JanusGraph documentation: a graph modeled in TinkerGraph seems to be be compatible with JanusGraph's graph model. Would my understanding be correct? Our aim is to create and store such a graph on a low end device and then transfer the data to an Apache Atlas installation running on a high end cloud server, the graph model of Apache Atlas being built on JanusGraph.

[Jason Plurad]

Yes, both TinkerGraph and JanusGraph implement the same Apache TinkerPop APIs.
I'm not familiar to what extent Apache Atlas exposes its underlying graph database, but all of those projects have TinkerPop in common.