On my web-app, the user can request different data lines. Those data lines have an unique "statusID" each, say "18133". When the user requests to load the data, it is either loaded from the server or from indexedDB(that part im trying to figure out). In order to make it as fast as possible, I want the index to be the timestamp of the data, as I will request ranges which are smaller than the actual data in the indexedDB. However, I am trying to figure out how to create the stores and store the data properly. I tried to dynamically create stores everytime data with a new Id is requested, but creating stores is only possible in "onupradeneeded". I also thought about storing everything in the same store, but I fear that the performance will make that bad. I do not really now how to approach this thing.
What I do know: If you index a value, it means that the data is sorted, which is exactly what I want. I dont know if the following is possible but this would solve my issue too: store everthing in the same store, index by "statusID" and index by "timestamp". This way, it would be fast too i guess.
Note that I am talking about many many datapoints, possible in the millions.
You can index by multiple values, allowing you to get all by statusID and restricting to a range for your timestamp. So I'd go with the one datastore solution. Performance should not be an issue.
This earlier post may be helpful: Javascript: Searching indexeddb using multiple indexes