Note: Some of these are close to complete or in-progress, but we have not yet organized this brainstorm list:
Many of these are already complete but we should include all popular / desired features.
- Internal Metrics - Track internal operational stats so that we can eat our own scalability-improving dog food.
- Annotations - visual notes for events which occur concurrently with the data.
AVG,MIN,MAX,SUMaggregate functions across data.
- stacking of multiple metrics
AREA/LINEgraphs (related to stack)
- zoom into a graph within a specific range.
- shifting time ranges. Same time interval but jump back in time.
- roles/groups of metrics. These are tags for specific metrics.
- bounds for metrics (min, max)
NaNgraphed as a different value (grey).
- don't show metrics when selecting a role / cluster and that metric isn't on a specific machine.
- groups for metrics (
- scale to LOTS of metrics.... millions per collector
- title+description support for
smetricso that these strings show in the UI (metadata) Maybe also key value pairs on top of the metrics. Category / tag would be another set of metadata.
- protocol for receiving metrics in text, json, XML, protocol buffers, etc.
gangliaso that if a metric is no longer used it ommitted from the UI. This might just be metadata.
- named values for time ranges (hour, day, week, month, etc).
- configured graph sizes by name (large, large1, large2, medium, etc).
- Permalinks - after you zoom into a page, it becomes permalinkable like google maps / google finance (link to this chart).
- configuration of the number of replicas
- multiple datacenters
- multiple datacenter sync on commit