The biggest updates for Heroic are highlighted below. For a full list of changes, see the relevant release on GitHub.
Changes listed for future releases are expectations but not hard commitments. The roadmap may be updated at any time.
Theme: Metric storage improvements.
- BREAKING Text-based serialization for BigTable RowKeys, allowing for better integration with standard BigTable tools.
- Retention policy automatically set on BigTable clusters.
- New timeseries sampled and derived from raw data upon ingestion.
- Configurable options for BigTable partitioning.
Theme: Advanced query usability.
- Aggregations for rate of change.
- Processing of arithmetic queries.
- Query clients for Java and Python.
- Better support for querying with a non-JSON DSL.
Theme: Stability and documentation.
- BREAKING Removal of support for events. Attempting to send events (or any non-metric data) through the ingestion pipeline will generate an error.
- Documentation overhaul, complete with architectural references.
- Java 11 compatability.
- Analytics added on startup to track deployments and usage. Analytics can be disabled with a config option.
- BREAKING Update semantic-metrics for internal instrumention, replacing meters with counters.
- Use protobuf for shell serialization.
- BREAKING Remove deprecated key-value v1 metadata backend. ElasticSearch is still fully supported as the MetadataBackendKV module.
- Enable dynamic ElasticSearch node lookup.