I read in a manual that about the Qlikview memoyr usage. The understanding that I got is that Qlikview will not optimize memory consumption until it reaches the minimum working set limit. That means that if my server is using memory till working set limit min, some of the memory might be used for cache and some of the memory might be used for actual document usage
After the working set min is reached, the document will start deleting the old cache to make space for new documents for caches
Now my questions is whether there is any way to actually tell how much actual memory is in usage so that we can know when is the time when we might need to buy more ram. Our servers often stay at server's min working set limit so it is hard for us to tell whether we still have enough memory that is in use for old cache or whether all the memory is being used by the document and now we should buy more memory
If you servers stay at the minimum working set limit, I would say that it may be time to start reflecting on a possible RAM purchase. The server RAM load is not at its maximum, because you still haven't reached the max working set limit. When your servers stay at the maximum limit, you should put on your coat and hurry to your local store to buy some more RAM.
Is it not so that Qlikview will not delete old cache/ optimize the memeory and will keep piling up the old cache until it reaches the working set min. That means that we might still have a lot of memory left and it is only being used by old cache. Qlikview will only start deleting old cache once it has going beyong working set min.
In this case, buying a new set of ram would be a waste if we will have a lot of memory left but Qlikview has not optimized it
Well, I was only suggesting to buy RAM when the server reaches the max working sel limit and stays there. Two commonly occurring events can influence memory usage in a dramatic way:
adding a pack of users to one or more documents
publishing new documents with a considerable footprint
These may seem to have predictable impact with respect to memory usage, but since QVS memory usage per document isn't really a transparent affair (not taking into account that QDS may add a considerable uncertainty factor if you run both on the same machine), making correct calculations may be close to impossible.
I think the best thing you can do is monitor memory usage and do trend analysis on a regular basis.