Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Renting Edge Computing Resources for Service Hosting

Aadesh Madnaik, Sharayu Moharir, Nikhil Karamchandani

TL;DR

A deterministic online policy is proposed and its performance for adversarial and stochastic i.i.d. request arrival processes is characterized to conclude that the policy is robust to temporal changes in the intensity of request arrivals.

Abstract

We consider the setting where a service is hosted on a third-party edge server deployed close to the users and a cloud server at a greater distance from the users. Due to the proximity of the edge servers to the users, requests can be served at the edge with low latency. However, as the computation resources at the edge are limited, some requests must be routed to the cloud for service and incur high latency. The system's overall performance depends on the rent cost incurred to use the edge server, the latency experienced by the users, and the cost incurred to change the amount of edge computation resources rented over time. The algorithmic challenge is to determine the amount of edge computation power to rent over time. We propose a deterministic online policy and characterize its performance for adversarial and stochastic i.i.d. request arrival processes. We also characterize a fundamental bound on the performance of any deterministic online policy. Further, we compare the performance of our policy with suitably modified versions of existing policies to conclude that our policy is robust to temporal changes in the intensity of request arrivals.

Renting Edge Computing Resources for Service Hosting

TL;DR

A deterministic online policy is proposed and its performance for adversarial and stochastic i.i.d. request arrival processes is characterized to conclude that the policy is robust to temporal changes in the intensity of request arrivals.

Abstract

We consider the setting where a service is hosted on a third-party edge server deployed close to the users and a cloud server at a greater distance from the users. Due to the proximity of the edge servers to the users, requests can be served at the edge with low latency. However, as the computation resources at the edge are limited, some requests must be routed to the cloud for service and incur high latency. The system's overall performance depends on the rent cost incurred to use the edge server, the latency experienced by the users, and the cost incurred to change the amount of edge computation resources rented over time. The algorithmic challenge is to determine the amount of edge computation power to rent over time. We propose a deterministic online policy and characterize its performance for adversarial and stochastic i.i.d. request arrival processes. We also characterize a fundamental bound on the performance of any deterministic online policy. Further, we compare the performance of our policy with suitably modified versions of existing policies to conclude that our policy is robust to temporal changes in the intensity of request arrivals.
Paper Structure (22 sections, 18 theorems, 24 equations, 8 figures, 1 table, 3 algorithms)

This paper contains 22 sections, 18 theorems, 24 equations, 8 figures, 1 table, 3 algorithms.

Table of Contents

  1. Introduction
  2. Our Contributions
  3. Related Work
  4. Setting
  5. Cost Model and Constraints
  6. Performance metrics
  7. Policies
  8. Better Late than Never (BLTN)
  9. Follow The Perturbed Leader (FTPL)
  10. An illustration
  11. Main Results and Discussion
  12. Simulations
  13. Conclusion
  14. BLTN Naive Algorithm
  15. Proof Outlines
  16. ...and 7 more sections

Key Result

theorem thmcountertheorem

Let $\Delta \kappa = \kappa_H - \kappa_L$, $\Delta c = c_H - c_L$, $W = W_{LH} + W_{HL}$. If $\Delta \kappa > \Delta c$ then,

Figures (8)

  • Figure 1: Gilbert-Elliot Model as a Markov Chain
  • Figure 2: Performance of various policies as a function of the request arrival rate. For the GE model, we fix $\lambda_L = 300$ and vary $\lambda_H$.
  • Figure 3: Performance of various policies as a function of $\Delta \kappa$
  • Figure 4: Performance of various policies as a function of difference in rent costs $\Delta c$
  • Figure 5: Performance of various policies as a function of switch costs, $W = W_{HL} = W_{LH}$
  • ...and 3 more figures

Theorems & Definitions (37)

  • remark thmcounterremark
  • theorem thmcountertheorem
  • lemma thmcounterlemma
  • theorem thmcountertheorem
  • lemma thmcounterlemma
  • proof
  • lemma thmcounterlemma
  • proof
  • lemma thmcounterlemma
  • proof
  • ...and 27 more