-
-
Notifications
You must be signed in to change notification settings - Fork 520
Description
Hey, in our privacy focused app we use a flexsearch.
It works nice, but currently we run import/export to ensure persistence.
We cannot use any of implemented storages, because we use a transparent encryption for all data in our app.
Our policy is the unencrypted data may appears only in RAM. That literally mean that we cannot place any temporary data out of RAM, for example in temporary directory or in IDB, etc.
The storage that may be mounted sounds as interesting performance opportunity for us. But I can't find any docs about custom storages and have some questions like
- What a purpose of a persistent storages? I mean does it purposed to offload a part of the data and efficiently load in RAM only those pieces of data that is needed for current search?
- What's the persistence storage API? I've tried to explore the implementations, but for me is not obvious the minimal required API we have to implement to introduce our custom storage
- Why so large differences between persistence backends in comparison table? What rules we should account to implement the performant backend?
It would be very nice to introduce a new markdown file with explanation of that points. The custom storage would be very useful in case it could help us to offload some part of data from a RAM into an encrypted disk.