Make monasca-statsd compatible with Prometheus statsd-exporter and DataDog
As an OpenStack Python developer I want to have a push-metrics client that works with regular StatsD/Graphite, Monasca, Prometheus statsd-exporter and DataDog, so that
* my application does not depend a specific single monitoring solution / vendor
* my applications does not have to collect, store and serve metrics (e.g. from multiple threads or worker processes)
* my application will still serve metrics even when its request pipeline, connection pools, ... are exhausted or overloaded
* my application does not have to open a dedicated port for monitoring or provide a special authorization profile for monitoring users
As am SRE, I want to benefit from the variable frequency of statsd-metrics (no preaggregation) and obtain micro-histograms for timer-metrics that provide me with SLA-relevant quantiles and peaks.
The Monasca-StatsD client can be adjusted to DataDog format with limited effort: The dimensions need to be serialized differently.
Ihe Monasca Agent statsd daemon can be extended to calculate micro-histograms with limited effort. For the beginning, it can just treat histograms and timers as gauge before creating a simple in-memory timeseries capturing data for a single measurement interval to calculate quantiles, max, etc..
Blueprint information
- Status:
- Not started
- Approver:
- Roland Hochmuth
- Priority:
- Undefined
- Drafter:
- jobrs
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- New
- Series goal:
- None
- Implementation:
- Unknown
- Milestone target:
- None
- Started by
- Completed by