There is no leakage between datasets in Fuseki. Access control will
prevent external access across users. To have different host, it would
need a reverse proxy; in a single Fuseki it will be the dataset name
that separates the tenants. The tenants have to trust the server to do
the right thing.
(it takes a Jetty restart to add a new connector to listen on a newly
added port)
Even if it can be done, having separate Fuseki's is clearer to the
tenants that there data is separate.
One tenant does not impact the performance (no accidental DOS).
Personally, I'd go for separate containers until there is an identified
advantage of a combined system.
Fuseki main starts up pretty quickly.
Andy
On 21/05/2019 14:38, Martynas Jusevičius wrote:
Hi,
I have a use case for deploying triplestores dynamically. They should
provide SPARQL Protocol and GSP endpoints on unique URLs and isolate
the datasets from each other, while also having a minimal footprint.
I'm thinking of using Fuseki Docker image for this [1].
Would I need to run a new Fuseki container for each dataset, or would
it suffice to create new datasets dynamically and have them share the
same Fuseki instance?
Martynas
[1] https://github.com/AtomGraph/fuseki-docker