cynthia_Kshlerin answered on October 14th 19 at 10:33
I could be wrong, but I think you can do without the database. Keep them all in some convenient data structure, like a HashMap and serialize it from time to time to the disk. For example in a separate thread every 0.1 seconds. But of course all depends on the amount and soloistic dokumentov that you are going to store.
kattie_Keeling answered on October 14th 19 at 10:35
rodolfo_Reichert answered on October 14th 19 at 10:37
As for caching, it is a great thing — Ehcache. But this is purely a cache, i.e., a trivial Map, Ala memcached (though with bells and whistles and can embedded). It all depends on what features of the search attributes you need — if everything is transparent and clear, and it is possible to build simple indexes and to use it, it will be fast.
I can still see Neo4J — embedded is a graph-oriented database, a very good thing, if your application falls on graphs.
Emelia_Shanahan answered on October 14th 19 at 10:39
And I recommend OrientDB. Very cool thing: the more that can be oriented in the desired direction: either keyValue or Graph oriented, relational, etc. www.orientdb.org/orient-db.htm