bradmontgomery
commented
10 years ago Thanks for the suggestion. There's a couple of reasons why I'm hesitant about doing the work to dump metrics into your relational DB:
- Redis's persistance is configurable. You can make it save to disk on every update, which would be really slow, but you'd never lose any data. If you're really concerned about volatility, this is probably the best area you need to investigate first.
- How would you want to store an export of your metrics? A snapshot of all your metric data (as json? binary data?)? A similar key/value schema (PostgreSQL has hstore, which could conceivably replace redis for an app like this, but that's a whole 'nother story)?
- I know this isn't what you've asked, but storing all metrics in your DB could really degrade performance (unless you've got a dedicated DB just for metrics). That's probably enough work to warrant a completely different project or whole refactor to allow different backends for this app... something I'd decided against doing from the get-go. If you want to go this route, take a look at
django-app-metrics.
Let me know your thoughts on the above, and maybe a little info about how your using this app, and we'll go from there!
At the moment the metrics are very volatile, as all data is stored only in the Redis server. It would be useful to be able to dump and load the metric data to the django DB. Probably not that trivial though...