openlink
What is the maximum amount of graphs
Assuming graph sizes vary between 100 and 1000 triples. Is there information available on how Virtuoso performs when using a large amount of graphs (between 100k and 1M)?
There is no theoretical limit on the number of graphs, nor on triples, given suitable increase of RAM and disk on a single-server instance. Optimal performance is generally seen when Virtuoso is allocated at least 10Bytes RAM per triple or quad, so given 10GB RAM, Virtuoso should handle roughly 1 Billion triples or quads.
You can see a large single-server deployment handling UniProt, with —
There are 111,761,129,065 triples in this release (2023_01). The query timeout is 45 minutes. All triples are available in the default graph. There are 15 named graphs.
There's also a Google Sheet with statistics for a number of public endpoints.
How Virtuoso or any other DBMS "performs" in any given situation is highly variable. How much RAM, what CPU power is available, what kinds of queries are being run, etc. The Linked Data Benchmark Council has some posts about benchmarks run on Virtuoso, which may answer some of your questions...
If you need more info, this might be a better question for our Community Forum, especially if you can provide more detail about your proposed scenario.
Thank you very much for the information.
The example has 15 names graphs, so the ratio triples per graph is very high (7bio/1). I’m interested to see if the performance remains the same if that ratio drops substantially (1000/1).
I’ll open a question in the community forum.