Feature #2492
Updated by Alexander Watzinger 10 months ago
With growing datasets, performance improvements are an ongoing task. When testing with "THANADOS":https://thanados.net which has a dataset of about 200,000 entities and 700,000 links, we noticed that loading the base system could already take a few seconds so we decided to focus on that. Measurements noted here are taken from a locally installed THANADOS instance. *To do* * Types (0.3 sec, 1.5 sec with count) ** Are loaded with count at entity view. But I think we only need it for type views. Maybe change the approach to add it where needed in contrast to try to figure it out early in the init script. * Reference systems (0.17 sec) * Init (0.14 sec) * Files (0.11 sec) but without having the actual files (should be tested with files) *Done* (already in the develop branch) * Skipping model counts where not needed -> *improvement improvement to 0.02 seconds instead 1.00 second* * Skipping type counts at entity default view for none types -> *improvement over 1 second when viewing (non type) entities* * Skipping model loading for image and logo display -> *improvement improvement varying but noticeable* noticeable * Tested performance on THANADOS server in comparison which looked ok