I would like to know the best practice in terms of stats for Redis. I would like to know what are the stats, but per prefixed keys.

I read it's possible to do that if we create multiple DBs, but in a cluster mode, it's not possible (or at least on AWS). Would you have a suggestion other than putting this responsibility on the client side?

Comment From: itamarhaber

Hello @elenigen

Keep in mind that this issue tracker should be used for reporting bugs or proposing improvements to the Redis server. Kindly close this.

Questions should be directed to the community:

Comment From: elenigen

To my understanding, this feature is missing, so it should be considered as an improvement proposal ... unless proven wrong. In other words, the Redis cluster should setup should allow the notion of DBs, so this way, we could have stats per key types.

Comment From: itamarhaber

If the feature request is to have multiple DBs in the cluster then the issue's title needs adjustment :) FYI, that omission is intentional and recorded in the cluster spec (https://redis.io/topics/cluster-spec#implemented-subset).

I regarded this as a question due to the use case provided the language used ("Would you have a suggestion other than putting this responsibility on the client side?"). If this is a feature request, it would be best if you could add more details about it, for example the types of statistics you're interested in. Is it only expired/dbsize/with ttl like shown in INFO? Is it also hits/misses per prefix? What else?

Finally, I had referred you to the community because someone there may propose ways of generating the data you're seeking (e.g. perhaps a proxy of sorts, an analysis of monitor/replication or similar) other than client-side tracking.

Comment From: itamarhaber

This issue will now be closed - please feel free to reopen if needed.