You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It makes sense to start the Neo4j database server on demand when Graphryder Dashboard is accessed, rather than keeping it running constantly. Because:
Neo4j has rather huge memory requirements (to be checked in more detail though).
We and most future users of this software only need Neo4j for Graphryder, and Graphryder is not accessed that often, so the whole main memory used for it would be wasted 99% of the day.
If each Graphryder API instance spins up and down its own associated Neo4j instance, we also solve the aggravated memory consumption problem of (right now) having to run multiple Neo4j instances, one per dataset. But that's a rather hacky way to "solve" it, and it's better properly done by making Graphryder API aware or capable of multiple datasets (tracked in Make this software aware or capable of multiple datasets in Neo4j #22 ).
It makes sense to start the Neo4j database server on demand when Graphryder Dashboard is accessed, rather than keeping it running constantly. Because:
Neo4j has rather huge memory requirements (to be checked in more detail though).
We and most future users of this software only need Neo4j for Graphryder, and Graphryder is not accessed that often, so the whole main memory used for it would be wasted 99% of the day.
If each Graphryder API instance spins up and down its own associated Neo4j instance, we also solve the aggravated memory consumption problem of (right now) having to run multiple Neo4j instances, one per dataset. But that's a rather hacky way to "solve" it, and it's better properly done by making Graphryder API aware or capable of multiple datasets (tracked in Make this software aware or capable of multiple datasets in Neo4j #22 ).