On 13/05/2024 11:10, Martynas Jusevičius wrote:
Hi,
I'm using multiple Fuseki instances in a Docker setup but considering
to use a single instance with multiple datasets instead.
So I was wondering what the differences of those setups are (besides
the lower memory consumption etc.)
which are not so great because the dominant memory cost is the database.
in terms of:
- security - I suppose there would be no difference since the datasets
are isolated and have separate endpoints?
If the docker setup is multi-machine, there is isolation of
denial-of-service issues.
- federation - would SPARQL federation perform better on a single
instance? E.g. if a query federates between datasets on the same
instance, maybe Fuseki would recognize that and avoid HTTP calls? Just
thinking out loud here.
- any other aspects?
Administration convenience - which could go either way.
Load balancing.
Martynas
Andy
