i have read some discussion topic on how to tune-up logstash configuration, and for what i have understand there are a bunch of variables and metrics that should be related.
- average shard's size should be between 20GB and 40GB
- the total number of shards inside a node should be 20*(GB-heap-size)
I wonder if, fixing some assertions ( ex: indexes are daily-time-based ), exists some math formula that consent to estimate the optimal shards number compatible with input parameter ( number of nodes, GB-heap-size for nodes, average daily indexes size, total number of indexes created daily, retention period for an index ).
I'd like to have some automatic computation that could help me to visualize different tuning option corresponding to different input parameters value .
Just some starting formula to have some rapid feedback, i'm not involved in modeling machine-learning auto-predicting models ) .
Thank u all for the feedback