Resource Allocation Algorithms and Strategies in Cloud Environments

October 13, 2021

Cloud computing is becoming more popular among small, medium, and large enterprises. These enterprises benefit from increased efficiency, cost optimization, data security, unlimited storage capacity, and more.

Yet the cost of cloud resources has been a common concern among these companies. This is why cloud resource optimization is a topic of discussion. Several methods optimize cloud computing resources to meet the actual demand and reduce cloud services costs.

This article digs deeper into resource allocation strategies in cloud computing and discusses resource allocation limitations in the cloud.

Table of contents

Prerequisites

To understand this article better, you’ll need to have a basic understanding of cloud computing. Please have a look at the introduction to cloud computing here.

Resource allocation strategies in the cloud

Cloud computing provides businesses, clients, and projects a computing environment for leasing resources on-demand. Both cloud consumers and suppliers have a goal to allocate cloud resources effectively and achieve profit.

These resources are often scarce, and cloud providers must leverage and distribute the resources within the limits of the cloud environment and meet the needs of cloud applications so that they complete their intended tasks.

One of the most critical cloud computing components is the allocation of resources. Its efficiency directly influences the performance of the whole cloud environment.

Resource allocation faces major challenges, including cost efficiency, response time, reallocation, computational performance, and scheduling tasks. Consumers of cloud computing services target to accomplish tasks with the lowest costs possible.

Resource allocation involves the provision of services and storage space for particular tasks given by users. This is accomplished using different resource allocation strategies.

Resource allocation strategies involve integrating cloud provider activities when allocating and utilizing scarce cloud resources, and meeting cloud applications’ needs so that they complete their intended tasks.

Cloud users and cloud providers, the two players in a cloud computing environment, have different goals to pursue. Cloud providers promote high resource utilization to maximize profits while users pursue an opposite goal. They want to cut down on cloud computing expenses without compromising their performance requirements.

There are different strategies to strike a balance between resource allocation and resource cost.

These strategies help to avoid:

  • Over-provisioning – this occurs when the available cloud resources exceed the demanded resources.
  • Under-provisioning – this happens when the allocated resources fall short of the demanded resources.
  • Resource fragmentation – this issue exists when the resources in the system are inaccessible. The resources available are not capable of assigning themselves to the required users.
  • Resource contention – this is a condition where two or more applications in the cloud system intend to use the same computing resources in the same instance.

In this article, we will discuss the following four resource allocation strategies:

  1. Dynamic resource allocation.
  2. Linear scheduling strategy.
  3. Particle swarm optimization.
  4. Ant colony optimization (ACO) algorithm.

Dynamic resource allocation

Dynamic resource allocation is used for load balancing, and it is a significant optimization process in cloud computing. This strategy distributes loads among virtual machines (VMs) to achieve a cost-efficient usage of computing resources. It emphasizes on-demand services and resource scalability.

Businesses scale their cloud resources dynamically based on their needs. While on-demand resource allocation may face challenges, VM technology is employed to provision resources.

Using a virtualized environment reduces the average job response time and ensures tasks are executed according to resource availability. Utilizing cloud resources dynamically is the key to balancing the load and avoiding issues such as the slow running of systems.

Linear scheduling strategy

When resource allocation and resource consumption are integrated, resource utilization is improved. Linear scheduling strategy seeks to allocate cloud resources while maximizing the system throughput as well as resource use. This technique distributes resources among the requesting users to maximize the Quality of Service (QoS) parameters such as throughput, response time, and service cost.

Linear strategy algorithms compile user requests and sort them in different queues as per their threshold value. The scheduling algorithm maximizes resource utilization by considering the available virtual machines and tasks.

Resource utilization in cloud environments is based on memory usage, CPU usage, and throughput. Thus, the scheduling algorithm uses a server node to determine the most applicable virtualization method.

An IaaS cloud environment to allocate resources that maximize resource usage and throughput. Linear strategy improves response time and reduces waiting time as opposed to scheduling tasks and resources separately.

Ant colony optimization (ACO) algorithm

ACO is an optimization algorithm inspired by techniques used by some ant species. At each interaction of ACO, these ants deposit an artificial hormone (pheromone) on the ground that acts as a mark to the best possible path for other colony members to follow. This is a powerful scheduling method in cloud computing.

There are several challenges associated with scheduling in the cloud computing infrastructure. Examples include load balancing, budget, and computation time issues. ACO is a significant approach to solving load balancing problems. This approach achieves better resource usage and higher throughput.

It does so by balancing the load in the cloud infrastructure and minimizing the computational time or makespan. In cloud computing, makespan represents the completion time of the last task to leave the system. ACO manages the usage of virtual machines, memory, and the number of clouds.

Particle swarm optimization algorithm

The particle swarm optimization (PSO) algorithm is made up of a set of particles characterized by positions and velocities. The velocity in a PSO particle represents the movement in the search space, which adjusts dynamically based on previous behavior. Particles continuously move within the search space in search of better points. They change their velocity and position to arrive at the solution space.

PSO method is used in complex, large-scale environments. Its performance in reducing computational time is unmatched with other swarm intelligence algorithms, including genetic algorithms (GA). PSO uses absolute values of numbers, unlike the GA algorithms that encode values to binary.

PSO provides effective optimization in cloud computing and other distributed environments. It is faster in computation time compared to ACO, GA, and other meta-heuristic algorithms. Due to its simple processing and faster processing time, PSO is a reliable solution to the task scheduling problem.

Limitations of resource allocation in the cloud

  • Cloud users rent resources from remote servers but do not have control over their resources. Thus, migration to other cloud providers, searching for better data storage, and other services alike is challenging.
  • Transfer of data from one provider to another may pose risks. Data stored in the public cloud is susceptible to phishing and hacking attacks. Malware can spread quickly because servers on the cloud are interconnected.
  • More profound knowledge is required to master resource allocation and management. This is so because the cloud service providers hold a lot of information about the working of the cloud.
  • You’ll need to install software locally if you’re targeting to use peripheral devices such as scanners and printers with the cloud. Networked peripherals work well with the cloud.

Conclusion

Enterprises continue to rely on cloud computing technology for many use cases, including increasing efficiency, optimizing cloud costs, ensuring data security, and storing unlimited data. Resource allocation knowledge is a growing need for cloud users who want to optimize their cloud usage costs.

While there are many strategies for allocating resources in the cloud, the most effective method is the one that achieves cloud user satisfaction and maximizes the income for cloud service provers.

Happy learning!

Further reading


Peer Review Contributions by: Willies Ogola


About the author

Jayden Kiprotich

Jayden is a Computer Engineering undergraduate student. He enjoys writing software and firmware for embedded microcontrollers. He beliefs in the power of teamwork to build a sense of comradery as engineers work towards achieving a common goal.

This article was contributed by a student member of Section's Engineering Education Program. Please report any errors or innaccuracies to enged@section.io.