Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Seer: Proactive Revenue-Aware Scheduling for Live Streaming Services in Crowdsourced Cloud-Edge Platforms

Shaoyuan Huang, Zheng Wang, Zhongtian Zhang, Heng Zhang, Xiaofei Wang, Wenyu Wang

TL;DR

This work addresses the challenge of maximizing CCP revenue in crowdsourced cloud-edge platforms amid a nonlinear link between resource utilization and income. It introduces Seer, a proactive revenue-aware scheduling system implementing a Pre-schedule-Execute-Re-schedule (PER) paradigm that couples spatio-temporal request prediction with a revenue-driven optimization, then executes and adjusts in real time. Key contributions include the AE-GRU-based spatial-temporal prediction, XGBoost-based revenue modeling, a tractable LP-based pre-scheduling solver with dimensionality reduction, and flexible Conservative vs. Aggressive modes to trade off revenue and reliability. On real CCP data, Seer delivers up to 147% higher revenue than an Origin baseline, nearly 4× gains over a GP baseline, and up to 3.4× faster scheduling, demonstrating significant practical impact for large-scale live streaming services.

Abstract

As live streaming services skyrocket, Crowdsourced Cloud-edge service Platforms (CCPs) have surfaced as pivotal intermediaries catering to the mounting demand. Despite the role of stream scheduling to CCPs' Quality of Service (QoS) and throughput, conventional optimization strategies struggle to enhancing CCPs' revenue, primarily due to the intricate relationship between resource utilization and revenue. Additionally, the substantial scale of CCPs magnifies the difficulties of time-intensive scheduling. To tackle these challenges, we propose Seer, a proactive revenue-aware scheduling system for live streaming services in CCPs. The design of Seer is motivated by meticulous measurements of real-world CCPs environments, which allows us to achieve accurate revenue modeling and overcome three key obstacles that hinder the integration of prediction and optimal scheduling. Utilizing an innovative Pre-schedule-Execute-Re-schedule paradigm and flexible scheduling modes, Seer achieves efficient revenue-optimized scheduling in CCPs. Extensive evaluations demonstrate Seer's superiority over competitors in terms of revenue, utilization, and anomaly penalty mitigation, boosting CCPs revenue by 147% and expediting scheduling $3.4 \times$ faster.

Seer: Proactive Revenue-Aware Scheduling for Live Streaming Services in Crowdsourced Cloud-Edge Platforms

TL;DR

This work addresses the challenge of maximizing CCP revenue in crowdsourced cloud-edge platforms amid a nonlinear link between resource utilization and income. It introduces Seer, a proactive revenue-aware scheduling system implementing a Pre-schedule-Execute-Re-schedule (PER) paradigm that couples spatio-temporal request prediction with a revenue-driven optimization, then executes and adjusts in real time. Key contributions include the AE-GRU-based spatial-temporal prediction, XGBoost-based revenue modeling, a tractable LP-based pre-scheduling solver with dimensionality reduction, and flexible Conservative vs. Aggressive modes to trade off revenue and reliability. On real CCP data, Seer delivers up to 147% higher revenue than an Origin baseline, nearly 4× gains over a GP baseline, and up to 3.4× faster scheduling, demonstrating significant practical impact for large-scale live streaming services.

Abstract

As live streaming services skyrocket, Crowdsourced Cloud-edge service Platforms (CCPs) have surfaced as pivotal intermediaries catering to the mounting demand. Despite the role of stream scheduling to CCPs' Quality of Service (QoS) and throughput, conventional optimization strategies struggle to enhancing CCPs' revenue, primarily due to the intricate relationship between resource utilization and revenue. Additionally, the substantial scale of CCPs magnifies the difficulties of time-intensive scheduling. To tackle these challenges, we propose Seer, a proactive revenue-aware scheduling system for live streaming services in CCPs. The design of Seer is motivated by meticulous measurements of real-world CCPs environments, which allows us to achieve accurate revenue modeling and overcome three key obstacles that hinder the integration of prediction and optimal scheduling. Utilizing an innovative Pre-schedule-Execute-Re-schedule paradigm and flexible scheduling modes, Seer achieves efficient revenue-optimized scheduling in CCPs. Extensive evaluations demonstrate Seer's superiority over competitors in terms of revenue, utilization, and anomaly penalty mitigation, boosting CCPs revenue by 147% and expediting scheduling faster.
Paper Structure (27 sections, 10 equations, 15 figures)

This paper contains 27 sections, 10 equations, 15 figures.

Table of Contents

  1. Introduction
  2. related work
  3. MEASUREMENTS AND MOTIVATIONS
  4. Dataset Description and Features Describing
  5. Potential Improvement and Optimal Revenue Range
  6. Spatial-Temporal Correlation of Requests
  7. Principal Factors Affecting Request Revenue
  8. SYSTEM OVERVIEW
  9. Proactive Pre-scheduling Stage
  10. Request Prediction Model
  11. Modeling Request Revenue
  12. Pre-scheduling Model
  13. Pre-strategy Execution Stage
  14. Re-scheduling Stage
  15. Flexible Scheduling Model
  16. ...and 12 more sections

Figures (15)

  • Figure 1: Crowdsourced Cloud-edge service Platform.
  • Figure 2: Server utilization analysis.
  • Figure 3: Fitting plot of server utilization versus revenue.
  • Figure 4: Temporal fluctuation and autocorelation analysis.
  • Figure 5: Spatial correlation analysis.
  • ...and 10 more figures