Sharing and Caring of Data at the Edge
summary
Argues for a first-class edge data store and enumerates eight design requirements (abstractions, locality, heterogeneity, mobility, failover, scalability, semantics, monitoring). Survey finds most proposals meet locality and scalability but few handle heterogeneity or mobility. Proposes Griffin, a prototype that implements a multi-consistency declarative API, uses graph-based resource management and infrastructure-centric monitoring with mobility prediction for proactive replication and placement; prototype is in development and the authors solicit feedback on usability, management and incentive models.
AI-written summary · about
chapters (5)
- 0:00Start
- 4:26Design Requirements for Edge Store
- 5:20How To Ensure Smooth Session Migrations Failover
- 6:17Analysis of Existing S-Store Proposals
- 9:04Usability
related talks
25mAccelerating Data Processing and Consumption for Building Modern Apps at the Edge
29mEnhancing Flexibility and Productivity with Access Patterns and Storage-Agnostic Abstractions
21mThe Decentralized Data Access Revolution
01h 00mData Mesh: Beyond the Data Lake
55mThe Great Data Debate 2022
36mIntroducing Data Mesh A Paradigm Shift In Analytical Data Management Architecture
14mSelective Reliability Engineering: There Is No Single Source...
07mData Gravity and Kubernetes: Managing Large-Scale Data Ingest With Minimal Latency