As we needed some kind of ID cache for synchronization purposes, which is fast and durable. By having some experience with CouchDB, a NoSQL solution seems to be worth a try. I decided to do some performance tests against the already available and running MongoDB in our server infrastructure.
Formerly we filled a LinkedList (yes, it is a bigger Java project as the "java" tag suggests ;)) by an expensive SQL database at server start (Oracle or SQL Server). Afterwards we had to check, again very expensive, with "contains" if elements are inside. Also the list was limited to a fixed size and contained only the newest IDs.
Formerly we filled a LinkedList (yes, it is a bigger Java project as the "java" tag suggests ;)) by an expensive SQL database at server start (Oracle or SQL Server). Afterwards we had to check, again very expensive, with "contains" if elements are inside. Also the list was limited to a fixed size and contained only the newest IDs.
