Dumping ongoing results of the spike on top of various logging solutions.
#### h4. Prometheus
##### h5. Pros
- Shared service with backend metrics
- Requires an additional service for long-term metrics storage
##### h5. Cons
- Data model is time-series exclusive
- Pull-based data gathering
#### h4. ElasticSearch
##### h5. Pros
- High advertised performance
- Highest variety of data types and modelling (supports time series data and non-time series data)
- Rich querying possibilities
##### h5. Cons
- High minimum resource requirements (over evaluation-sized clusters)
#### h4. MongoDB
##### h5. Pros
- Supports time series data and non-time series data
##### h5. Cons
- Support for time series data is not native thus needs modeling work
- Grafana/Kibana cannot talk to MongoDB without ugly workarounds
- We can't go with the approach Prometheus storing data in MongoDB since it is not supported
#### h4. InfluxDB
##### h5. Pros
- Long-term storage adapter for Prometheus
- HTTP-based API for push-based metrics gathering
- Flexible querying with direct comparison to SQL
##### h5. Cons
- Focused on time-series data
- No clustering support in OSS offering
- Medium-sized minimum resource requirements |
|
