An Efficient Cloud Storage Model with Compacted Metadata Management for Performance...
summary
Implements a prototype called CloudTS that separates metadata from compressed time-series objects on Amazon S3, builds a global tag dictionary (Patricia-trie style) and a compact timeseries–tag mapping using a two-dimension bitmap/bit-packed sparse matrix, and groups related series into TS objects by tag similarity. Evaluation against Cortex, Apache Parquet and JSON Time Series on S3 shows a real prototype speedup (1.37x average, 1.43x in production runs), up to 36% lower latency and ~30% less data transferred.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:44Background and motivation
- 3:26Data organization issues
- 6:20CloudTS architecture
- 8:16Indexing and data layout
- 11:20Experimental evaluation
- 13:48Conclusion
related talks
39mA New Data Model for Time Series Data Processing- 39mA New Data Model for Time Series Data Processing
36mAn Exploration in Storing Telemetry in Cloud Object Storage
24mTSDB Data Compactions on Scale Without Headache: Now & Future
39mDeep Dive: Cortex
13mTransform your business with time-series data (PRT061)
24mCracking the Cost Code: A Cost Optimized Approach to Distributed...
30mScaling Time-Series Data to Infinity: A Kubernetes-Powered Solution with Envoy