I'm working on a project (embedded C++) which collect a lot of technical stats and store it (historical choice) in an sqlite3 local database. We have no need of relational db. The next evolution should be to allow the stored data to be retrieved with web services (REST Api) so I think we would add a little Node JS entry point beside the C++ data collector. I'we already done something similar in the past using a mutual sqlite db shared by native collector and a minimal nodeJS server, but I wonder if using something like mongodb could be a better idea.
The pros/cons I see :
pros :
- no sql. Actually we don't need id and the actual code becomes more and more complex because of it (compared to what we really need to do).
- easy to access from any process (no more read/write concurrent problems)
cons :
- need to have a mongo daemon running : is it a lightweight process ? (it will be embedded beside the collector process)
- over-sized for this usage
I also had a look at the UnQlite project it seems to be much more adapted to key/value storage (I think we need much more a document oriented solution).
Another important point : the solution should run on Linux and windows platforms the same way.
Thanks a lot for any advice or other suggestion.
Aucun commentaire:
Enregistrer un commentaire