Keyun Cheng

Sharing and Caring of Data at the Edge

Download

HotEdge, 2020

Summary

This paper studies edge computing, including the edge enviroment / architecture, edge applications, existing (storage) solutions, and challenges. This paper also brings an initial design of Griffin, a shared stoarge devices for the edge. A set of open topics are raised for discussion and follow-ups.

Details

Architecture

1. Features of edge
   • Distributed, Heterogeneous, dynamic architecture
2. Applications on edge
   • ML
   • Games
   • autonomous driving
   • Stream processing
   • Analytics
   • Caching
3. Challenges

• Expressive APIs
• Data Locality
• Impacts of heterogeneous environment
  • Load balancing
• Mobility of clients
• Failure protection
• Scalability
• Flexible consistency model for vaious applications
• Monitoring
  • Architecture, system status

Open issues (Picked from Sec 5 from the paper)

• How to do benchmark

• How to do resource provisioning on edge

• Data management

• How to efficiently monitor the edge and end devices

paper collections

introduction to edge computing

• The Emergence of Edge Computing.

• ETSI. Multi-access Edge Computing (MEC): Framework and Reference Architecture.

Systems

• (SEC’17) CloudPath: A Multi-Tier Cloud Computing Framework.

• (CAN’16, DataBox, proposal) Personal data management with the databox: What’s inside the box?

• (HotEdge’18, proposal) ECO: Harmonizing edge and cloud with ml/dl orchestration.

• (HotEdge’18) Datafog: Towards a holistic data management platform for the iot age at the network edge.

• Cassandra: a decentralized structured storage system.

General Architecture

• (HotEdge’19) An edge-based framework for cooperation in internet of things applications.

ML and DL

• (HotEdge’18) MODI: Mobile deep inference made efficient by edge computing.

• (HotEdge’18) esgd: Communication efficient distributed deep learning on the edge.

• Exploring the use of synthetic gradients for distributed deep learning across cloud and edge resources.

Videos

• (SEC’17) Parkmaster: an in-vehicle, edge-based video analytics service for detecting open parking spaces in urban environments.

• (SEC’18) Videoedge: Processing camera streams using hierarchical clusters.

• (SEC’18) Application-aware iot camera virtualization for video analytics edge computing.

• (SEC’18) Edgebox: Live edge video analytics for near real-time event detection.

Games

• Cloudfog: Leveraging fog to extend cloud gaming for thin-client mmog with high quality of service.

VR

• (SEC’17) Towards efficient edge cloud augmentation for virtual reality mmogs.

Stream Processing

• Distributed data stream processing and edge computing.

Collaboration

• Decaf: Iterative collaborative processing over the edge

Parameter Server

• (OSDI’14 Parameter Server) Scaling distributed machine learning with the parameter server.

Serverless computing

• The rise of serverless computing.

• Serverless is more: From paas to present cloud computing.

• (OSDI’18) Pocket: Elastic ephemeral storage for serverless analytics.

Database

• Edgedb: An efficient time-series database for edge computing.

• (TC’13 Google DB) Spanner: Google’s globally distributed database.

Techniques

• Incremental deployment and migration of geo-distributed situation awareness applications in the fog.

• Latency mitigation strategies.

• (TMC’18) Uloof: A user level online offloading framework for mobile edge computing.

• (WF-IoT’13) Geelytics: Geo-distributed edge analytics for large scale iot systems based on dynamic topology

• (INFOCOMM’18) Service entity placement for social virtual reality applications in edge computing.

Caching

• Precog: refetching for image recognition applications at the edge.

• Cachier: Edge-caching for recognition applications.

• (HotEdge’18) Towards a solution to the red wedding problem.

• (SIGMOD’18) Dpaxos: Managing data closer to users for low-latency and mobile applications

Load-balancing

• (SoCC’18) Fast and accurate load balancing for geo-distributed storage systems.

• (EDBT’18) Global-scale placement of transactional data stores.

Consistency

• (SOSP’13) Consistency-based service level agreements for cloud storage.

• Consistency models for cloud-based online games: The storage system’s perspective.

• Cloudcraft: Cloud-based data management for mmorpgs.

• (HotEdge’18) Toward session consistency for the edge.

• (DEBS’18) Fogstore: A geo-distributed key-value store guaranteeing low latency for strongly consistent access.

Synchronization

• (IoT’18) An architecture for iot clock synchronization.

Heterogeneous

• (IoTDI’19) Heteroedge: Taming the heterogeneity of edge computing system in social sensing.

Analysis

• The case for vm-based cloudlets in mobile computing.

• A survey of multi-access edge computing in 5g and beyond: Fundamentals, technology integration, and state-of-the-art.

• Performance Analysis of Object Store Systems in a Fog and Edge Computing Infrastructure

• (SEC’19) Why cloud applications are not ready for the edge (yet).

Benchmark

• (SoCC’10, Cloud YCSB) Benchmarking cloud serving systems with ycsb

• (SEC’17) Defog: Fog computing benchmarks.

• (SEC’18) Cavbench: A benchmark suite for connected and autonomous vehicles

Industry

1. Azure IoT Edge. Link

2. AWS IoT for the Edge. Link

3. Google Cloud IoT. Link

Strength

N/A

Weakness

N/A