Running objects and how many cached objects is too much
Have recently found the metric functions in sprintr (today) and we have been using them to peer into our test environment. We found that a certain app (PivotDataWidget) caches a lot objects when opened! If the page with the pivot widget is opened numerous times it keeps adding to the cached objects. Back to the real question - how many cached objects is too much? Does it really have that much of an effect on the system? I am assuming here that it is server based caching (Using the server's RAM to cache it), if so it would come down to the amount of RAM the server has. Is there a way to force a cache clear for a user or for a certain process? Lots of questions asked, but thanks for any replies!
We are currently experiencing some challenges with this and are trying various solutions. We saw that on a small node, 200.000 objects caused our server to crash and 50.000 objects made the experience pretty horrible for our users.
A straight-forward solution is to delete cache objects that are not needed. For example, when you create a filter object for a user, delete the previous one for that user.
However, with many users and complex filters, this could still lead to the 50.000 (as it did for us). We are currently building solutions that directly query the database (using OQL) and render it on the front end (using custom widgets). This brings down the cached objects to a minimum and works really well for us so far.