Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Power Aware Container Placement in Cloud Computing with Affinity and Cubic Power Model

Suvarthi Sarkar, Nandini Sharma, Akshat Mittal, Aryabartta Sahu

TL;DR

The paper tackles power-aware container placement in heterogeneous data centers by jointly considering multi-resource demands, affinity/anti-affinity constraints, and a cubic power model. It introduces PAP, AAP, and CPAAP as heuristic approaches, with CPAAP balancing low utilization and high affinity to minimize total system cost while maximizing affinity satisfaction. Empirical results on synthetic and Google cluster traces show CPAAP achieves up to 26% cost reduction and up to 37% affinity payoff improvement, outperforming GCCS and HOP-CAPM. This work demonstrates that integrating power-aware and affinity-driven strategies yields practical gains for scalable, energy-efficient container orchestration.

Abstract

Modern data centres are increasingly adopting containers to enhance power and performance efficiency. These data centres consist of multiple heterogeneous machines, each equipped with varying amounts of resources such as CPU, I/O, memory, and network bandwidth. Data centers rent their resources to applications, which demand different amounts of resources and execute on machines for extended durations if the machines provide the demanded resources to the applications. Certain applications run efficiently on specific machines, referred to as system affinity between applications and machines. In contrast, others are incompatible with specific machines, referred to as anti-affinity between applications and machines. We consider that there are multiple applications, and data centers need to execute as many applications as possible. Data centers incur electricity based on CPU usage due to the execution of applications, with the cost being proportional to the cube of the total CPU usage. It is a challenging problem to place applications on the machines they have an affinity for while keeping the electricity cost in check. Our work addresses the placement problem of matching applications to machines to minimize overall electricity costs while maximizing the number of affinity pairs of machines and applications. We propose three solution approaches: (a) Power-Aware Placement (PAP): applications are placed on machines where power usage is minimized, (b) Affinity-Aware Placement (AAP): applications are placed on machines where affinity is maximized, (c) Combined Power-Affinity Placement (CPAAP): this approach integrates the benefits of both PAP and AAP. Our proposed approach improves the affinity satisfaction ratio by up to 4% while reducing the total system cost by up to 26% and improving the affinity payoff ratio by up to 37% compared to state-of-the-art approaches for real-life datasets.

Power Aware Container Placement in Cloud Computing with Affinity and Cubic Power Model

TL;DR

The paper tackles power-aware container placement in heterogeneous data centers by jointly considering multi-resource demands, affinity/anti-affinity constraints, and a cubic power model. It introduces PAP, AAP, and CPAAP as heuristic approaches, with CPAAP balancing low utilization and high affinity to minimize total system cost while maximizing affinity satisfaction. Empirical results on synthetic and Google cluster traces show CPAAP achieves up to 26% cost reduction and up to 37% affinity payoff improvement, outperforming GCCS and HOP-CAPM. This work demonstrates that integrating power-aware and affinity-driven strategies yields practical gains for scalable, energy-efficient container orchestration.

Abstract

Modern data centres are increasingly adopting containers to enhance power and performance efficiency. These data centres consist of multiple heterogeneous machines, each equipped with varying amounts of resources such as CPU, I/O, memory, and network bandwidth. Data centers rent their resources to applications, which demand different amounts of resources and execute on machines for extended durations if the machines provide the demanded resources to the applications. Certain applications run efficiently on specific machines, referred to as system affinity between applications and machines. In contrast, others are incompatible with specific machines, referred to as anti-affinity between applications and machines. We consider that there are multiple applications, and data centers need to execute as many applications as possible. Data centers incur electricity based on CPU usage due to the execution of applications, with the cost being proportional to the cube of the total CPU usage. It is a challenging problem to place applications on the machines they have an affinity for while keeping the electricity cost in check. Our work addresses the placement problem of matching applications to machines to minimize overall electricity costs while maximizing the number of affinity pairs of machines and applications. We propose three solution approaches: (a) Power-Aware Placement (PAP): applications are placed on machines where power usage is minimized, (b) Affinity-Aware Placement (AAP): applications are placed on machines where affinity is maximized, (c) Combined Power-Affinity Placement (CPAAP): this approach integrates the benefits of both PAP and AAP. Our proposed approach improves the affinity satisfaction ratio by up to 4% while reducing the total system cost by up to 26% and improving the affinity payoff ratio by up to 37% compared to state-of-the-art approaches for real-life datasets.
Paper Structure (29 sections, 15 equations, 9 figures, 1 table, 4 algorithms)

This paper contains 29 sections, 15 equations, 9 figures, 1 table, 4 algorithms.

Table of Contents

  1. Introduction
  2. Related Works
  3. System Model
  4. Affinity and Anti-affinity matrices
  5. Power Consumption Model
  6. Problem Formulation
  7. Solution Approach
  8. Power Aware Container Placement (PAP)
  9. Affinity Aware Container Placement (AAP)
  10. Combine Power and Affinity Aware Container Placement (CPAAP)
  11. Computational Complexity
  12. Experimental Setup and Results
  13. Simulation Setup
  14. Synthetic Dataset
  15. Real-life Dataset
  16. ...and 14 more sections

Figures (9)

  • Figure 1: Relationship between containers and applications
  • Figure 2: Container Instance Placement Model. The applications are marked in different colours: $p_1$ with H, $p_i$ with H and $p_N$ with H. Each applications are filled with a cylinder indicating the resource availability and requirements: H indicates CPU, H indicates memory, H indicates I/O, H indicates B/W.
  • Figure 3: Approximation of cubic power model using two linear models. $\pi_j^T$ denotes the splitting points of two linear models.
  • Figure 4: Results of experiment with 175 applications and different number of machines on real-life dataset
  • Figure 5: Results of experiment with 20 applications and different number of machines on synthetic dataset
  • ...and 4 more figures