A recent major project needs more than the limit of 100 categories and 100 category combinations and in response to client requests, we have now increased these to 500 categories and 500 category combinations.
A number of other small improvements have been made.
Whilst we recommend the use of a 'remote' MySQL database server for huge projects, we are aware that there is an additional cost implication. Many users are satisfied with the 'flat file' system which uses a built-in database server called SQLite. This can handle huge databases too, but it is not intended for use with multiple simultaneous users, including distributed computing, where many computers are doing calculations of different parts of the model at the same time. This is because SQLite may load some index tables into memory and if different computers change the indexes at the same time, the two copies will become different, causing problems. With MySQL, the indexing is all done on the server computer, so the indexes should never get out of sync.
The remote database can have problems with 'latency' i.e. the response time of the network rather than the data rate in bits per second. NoiseMap does not send large amounts of data, but it does require each packet to be acknowledged. If this takes a long time (and here half a second is a very long time) data transmission become sluggish. The local database system shouldn't suffer from this.
To provide more options for users, from the release of NoiseMap 2.5.6 it is possible to convert backups Of MySQL databases (.nmbackup files) to local SQLite databases (.nmdb files) and vice-versa. However, we do not recommend doing regular round trips between the two systems because they store floating-point numbers differently and conversion can lose precision.
