LOV performance is very good since the results are cached on the server as well as on the client.

The cache on the server (done with the help of the application server through the use of an BMP EJB) is meant to reduce the time needed to answer a client request for an already requested LOV. That means that once a LOV is requested (first case), the next requests for the same LOV will be faster (second case) since the server will not retrieve the informations from the database but from its cache instead. The cache on the client is the biggest factor in the performance increase (this case). Once a LOV is in cache on the client side, we can save many server calls and the performances are way better (less than 1 ms in most cases).

To compare the performances, you can choose between a small or a large lov (this one returns the first name and the last name for 1000 individuals (so it's quite a large LOV)). Of course there are methods to resynchronize the LOV on the client and/or on the server to ensure the validity of the LOV (but most of the time the LOV are to be used for static or quasi static data (a list of countries, activities, customer types...)).