CC BY
5УДК 004.4
Sheinman Vered
bachelor's degree University of Haifa (Haifa, Israel)
KEY PERFORMANCE INDICATORS FOR EVALUATING SOFTWARE EFFICIENCY
IMPROVEMENTS THROUGH CONTAINER TECHNOLOGIES
Аннотация: this article explores how container technologies (CTech) enhance software efficiency by identifying measurable KPIs. It reviews CTech and their operational principles, analyzing their influence on performance, resource utilization, and response time in various systems. Special attention is given to defining and classifying key performance indicators (KPIs) for objectively assessing the benefits of CTech implementation. The article also examines tools and methods for evaluating software performance, including Prometheus, Grafana, and ELK Stack. The experience of implementing CTech in various industries is analyzed based on American companies' practices.
Ключевые слова: CTech, efficiency, software, KPI, performance, resource utilization, response time.
Introduction.
In recent years, container technologies (CTech) have significantly transformed approaches to software development, deployment, and management. A defining feature of these technologies is their ability to isolate applications within independent containers, enabling efficient resource utilization and expediting scaling processes. In the context of modern IT infrastructures, which demand high flexibility and rapid adaptability to changes, containerization enhances both the development process and the operational performance of software products.
One of the key challenges in implementing CTech is evaluating its impact on software performance. Despite the widespread adoption of containers, many
organizations encounter difficulties in identifying KPIs that accurately reflect the improvements achieved through these technologies. This article aims to examine the KPIs used to assess performance enhancements in software when transitioning to CTech.
Main part. CTech - overview and principles of operation.
Container technologies provide a method to isolate applications and their dependencies into separate, self-contained units called containers, ensuring consistent operation across different environments. Each container encapsulates everything required to run an application: libraries, dependencies, configuration files, and even the operating system. This ensures consistent and reliable application performance, regardless of the surrounding environment. Such features make containers an invaluable tool for software development, testing, and deployment across diverse infrastructures.
The history of CTech originated as a solution to the «works on my machine» problem - where applications that functioned correctly on one computer often failed to run properly on other devices due to differences in configuration and installed software. Initially, containerization was implemented through operating system features such as chroot in Unix-based systems, evolving later with the development of more advanced virtualization tools. A pivotal milestone in the evolution of CTech was the launch of the Docker project in 2013, which made containerization accessible and user-friendly for a broad range of users [1]. Docker introduced tools for creating, testing, and running containers, along with an ecosystem for their management. By 2023, Docker accounted for over 32% of the containerization technology market [2].
With the advancement of containerization, new technologies and platforms have emerged, significantly enhancing the potential for automating deployment and container management processes. One of the most prominent solutions for container orchestration is Kubernetes - a system developed by Google for automating the deployment, scaling, and management of containerized applications. Kubernetes provides robust tools for managing containers in distributed environments, ensuring
fault tolerance, automatic scaling, and streamlined application state management. According to Statista, Kubernetes held a 24,4% share of the container technology market in 2023. Furthermore, approximately 59% of respondents reported using containers for most or all production applications and business segments within their organization, as well as for the majority of development and deployment processes
[3].
Another notable platform is OpenShift, which is built on Kubernetes and extends its capabilities by offering enhanced security, integration with CI/CD processes, and streamlined application management. OpenShift is tailored for enterprise use, providing tools for more granular control over application deployments and additional services to support software development and operations.
Additionally, a wide range of tools and platforms support various aspects of containerization and management. Among container runtimes, solutions such as Containerd, Podman, and CRI-O stand out as alternatives to Docker, emphasizing lightweight design and seamless integration with modern orchestrators. For managing containers in distributed systems, orchestration tools like Docker Swarm, OpenShift, and Nomad are commonly used. These platforms facilitate the automation of deployment, scaling, and monitoring of containers, which is particularly critical in large-scale infrastructures [4].
Containerization has emerged as a revolutionary technology in software development and operations, enabling the creation of isolated, lightweight, and scalable environments for applications. It simplifies dependency management, ensures software agnostic, and enhances resource efficiency. Due to these distinct advantages, CTech are being widely adopted across both small enterprises and large corporate systems, providing flexibility and reliability to infrastructures (table 1).
Table 1. Key advantages of containerization [5].
Advantage Description
Agnostic Containers include the application and all its dependencies, allowing them to run in any environment.
Scalability The ability to quickly increase or decrease the number of containers to adapt to workload demands.
Resource efficiency Containers use fewer system resources compared to virtual machines, leveraging a shared OS.
Process isolation Containers are isolated from one another, preventing conflicts between applications.
Faster deployment Containers start in seconds, streamlining CI/CD processes and enabling rapid software updates.
Infrastructure flexibility Support for operation in both local systems and cloud environments.
Enhanced security Isolation reduces the risk of vulnerabilities affecting other containers or the host system.
An integral part of modern software development and operations is the adoption of CTech. Their versatility allows for a wide range of applications - from creating isolated environments for testing to deploying complex microservices architectures. Contemporary containerization tools enable developers to seamlessly adapt their applications to various infrastructures, whether on-premises servers or cloud platforms. This approach not only reduces infrastructure management costs but also enhances the reliability and flexibility of all operational processes.
Definition and classification of KPIs for software.
Evaluating the quality and performance of software relies heavily on KPIs. They provide measurable insights into how effectively an application operates and help identify areas requiring optimization. The selection of appropriate KPIs depends on the objectives set for the system and the specifics of its use. The primary classification of KPIs can be divided into thrcategories (figure 1).
KPI
system throughput
operation execution speed
stability under load
Figure 1. Categories of software efficiency.
Performance encompasses metrics such as system throughput, execution speed, and stability under load. The resource utilization category evaluates the amount of system resources consumed by an application, including memory, CPU time, and disk space. Response time measures how quickly the system responds to user requests, which is particularly critical for real-time applications. In traditional infrastructure, both standard metrics and specific indicators are utilized (table 2).
Table 2. Standard and specific metrics for software evaluation in traditional infrastructure.
Type of metric Description Examples of metrics
Standard metrics Evaluate overall systemlevel software performance. CPU load, memory usage, network throughput, network latency.
Specific metrics Assess application-specific performance characteristics. Page load time (for web applications), query execution time (for databases), number of user errors.
The application of KPIs in traditional infrastructure requires regular data collection and analysis. Monitoring and analytics tools are employed to evaluate current performance and predict changes under increased load. Accurate definition and classification of KPIs provide a foundation for effective software optimization and improved user satisfaction.
The impact of CTech on KPIs.
The performance and resource utilization in software systems are significantly influenced by CTech. One of the primary mechanisms driving this effect is application isolation, which enables more efficient allocation of resources across containers [6]. Each container operates as an independent unit, allowing for optimization of its performance while minimizing overhead on system resources. As a result, containerization enhances the utilization of CPU, memory, and other resources, as containers do not require full virtualization and can run on a shared operating system, thereby substantially reducing operational costs.
Evaluating metrics in containerized systems has distinct characteristics. For instance, the utilization of CPU, memory, disk space, and network resources tends to be more flexible and dynamic compared to traditional solutions. In containerized environments, each of these metrics is measured for individual containers, enabling precise identification of which applications or services are consuming the most resources. This facilitates a more granular analysis of system performance and allows for rapid responses to potential issues. Moreover, containerized systems support effective allocation and limitation of resources for each container, enabling more precise control over workload distribution.
The adoption of CTech has significantly transformed software deployment and update processes [7]. Containers streamline these processes through their lightweight nature and independence from underlying infrastructure. Application updates in containerized systems become faster and more predictable. Unlike traditional approaches, where updates often require extensive time for testing and deployment, containerization allows for the rapid and secure update of only the modified components of a system, minimizing risks and downtime. This also enhances metrics related to response time and system stability. Thus, containerization not only optimizes resource utilization but also makes software management and update processes more efficient and cost-effective.
Methods and tools for evaluating software efficiency in containers.
To effectively assess performance and resource utilization in containerized systems, specialized monitoring and analysis tools must be employed (table 3).
Table 3. Tools for evaluating software efficiency in containers [8, 9].
Tool Description Advantages
Prometheus A monitoring system that collects real-time metrics about the state of containers and applications. Allows flexible metric customization, integrates with various data sources, and supports long-term data storage.
Grafana A data visualization platform that works in conjunction with Prometheus and other monitoring tools. User-friendly interface for data visualization, support for multiple data sources, and extensive customization options.
ELK Stack A toolset (Elasticsearch, Logstash, Kibana) for collecting, storing, and analyzing logs. A powerful system for processing large volumes of data, enables real-time log analysis, and provides robust search tools.
In addition to widely adopted tools, there is a broad range of other solutions for monitoring and analyzing the efficiency of containerized applications. For example, Docker Stats provides basic capabilities for monitoring container resource usage within Docker, allowing for straightforward tracking of parameters such as CPU, memory, and network utilization. cAdvisor is a lightweight tool designed for container performance analysis, enabling real-time monitoring of system resource usage. More advanced platforms, such as New Relic and Datadog, support integration with containers and microservices, offering deep analytics and data visualization capabilities. These tools provide diverse approaches to monitoring and can be selected based on the specific tasks and infrastructure requirements.
Despite their extensive capabilities, tools for evaluating software efficiency in containers come with several limitations. For instance, while Prometheus and Grafana offer flexibility in metric collection and visualization, they can be complex to configure and maintain, particularly in large distributed systems that require
intricate setups and integration with multiple data sources. ELK Stack is a powerful tool for log processing, but its deployment demands significant computational resources and can be challenging to configure in complex infrastructures. Simpler solutions, such as Docker Stats and cAdvisor, have limited functionality, making them less suitable for monitoring large-scale or distributed systems where more comprehensive performance and resource analysis is required. Meanwhile, advanced platforms like New Relic and Datadog provide robust monitoring capabilities, but their cost can become a significant factor, especially for larger organizations with numerous containerized applications. These limitations should be carefully considered when selecting tools for monitoring and evaluating software efficiency in containers, depending on the scale and specific needs of the business.
Additionally, various testing strategies and performance data collection methods are actively employed to evaluate the performance of applications in containers. One approach involves load testing, where containers are subjected to varying levels of load to assess their resilience and performance under different conditions. Such testing should include an evaluation of factors such as response time, stability under high load, and resource management during intensive usage scenarios.
To collect performance data, specialized agents are often deployed within containers to gather information on system metrics such as CPU, memory, and network resource usage. An alternative approach involves monitoring containers at the orchestrator level, such as Kubernetes, which offers built-in tools for collecting metrics on containers and clusters.
Examples of implementing CTech to enhance software efficiency.
The widespread adoption of CTech across various industries has significantly improved software performance. A notable example is the use of containerization in the financial sector in the United States. For instance, JPMorgan Chase implemented CTech to enhance the scalability and flexibility of its infrastructure. By leveraging Docker and Kubernetes for deploying microservices applications, the company achieved substantial improvements in system response time - an essential
factor for delivering real-time financial services. Containerization notably reduced transaction processing delays while ensuring a high degrof fault tolerance and enabling automatic scaling during periods of peak demand. These advancements enhanced the stability of banking systems and improved customer satisfaction.
Another example of the successful adoption of CTech is Netflix, a leading provider of streaming services. Netflix leverages containers for the development and deployment of its applications, resulting in improved stability and accelerated release cycles for new features. By utilizing Kubernetes for container orchestration, the company has optimized its software deployment and update processes. Containerization has enabled Netflix to significantly reduce the number of system failures and enhance application response times, directly contributing to an improved user experience. This has positively impacted KPIs, such as response time and service stability, by reducing failures and increasing the speed of service responses to user requests.
For Uber, a prominent player in the technology sector, containerization has had a profound impact on resource utilization and the overall efficiency of operational processes. Uber employs containers to manage its microservices architecture, enabling optimized resource allocation and expediting the testing and deployment of new application versions. By leveraging Docker and Kubernetes, the company has reduced service deployment times from several hours to just a few minutes, significantly accelerating development and update cycles. Resource utilization metrics, such as CPU and memory, have also been improved, resulting in a substantial reduction in operational costs and enhanced overall infrastructure efficiency.
These examples demonstrate how the adoption of CTech can significantly enhance software development, testing, and deployment processes, while also improving the overall efficiency of IT infrastructure in large enterprises.
Conclusion.
Enhancing software efficiency has been greatly facilitated by the adoption of CTech, which offer significant advantages in resource management, scalability, and
deployment speed. Analyzing KPIs not only evaluates the impact of containerization on system performance but also identifies areas for optimization. Based on examples of successful container adoption across industries such as finance, streaming services, and the technology sector, it is evident that containerization improves response times, system stability, and reduces operational costs. However, to maximize the benefits of CTech, it is essential to consider the specific characteristics of the infrastructure and select appropriate monitoring and analysis tools. This approach ensures sustainable development of IT systems and their adaptation to modern business demands.
СПИСОК ЛИТЕРАТУРЫ:
1. Muzumdar P., Bhosale A., Basyal G. P., Kurian G. Navigating the Docker Ecosystem: A Comprehensive Taxonomy and Survey // arXiv preprint arXiv:2403.17940. 2024;
2. Leading containerization technologies market share worldwide in 2023 / Statista // URL: https://w.statista.com/statistics/1256245/containerization-technologies-software-market-share/;
3. Kubernetes - statistics & facts / Statista // URL: https://w.statista.com/topics/8409/kubernetes/;
4. Kuznetcov I.A. Optimization of distributed systems for mobile applications: improving performance and scalability // Innovative Science. 2024. № 5-1. P. 52-57;
5. Narasimhulu M., Mounika D. V., Varshini P., Amarenda K., Rao R. K. Investigating the Impact of Containerization on the Deployment Process in DevOps // 2023 2nd International Conference on Edge Computing and Applications (ICEC). I.E.E.E., 2023. P. 679-685;
6. Li H., Johra H., Pereira F. A., Hong T. Data-driven key performance indicators and datasets for building energy flexibility: A review and perspectives // Applied Energy. 2023. Vol. 343. P. 121217;
7. Sidorov D. Leveraging web components for scalable and maintainable development // Sciences of Europe. 2024. № 150. P. 87-89;
8. Malygin D. S. Monitoring the Availability of Web Services in Distributed Infocommunication Systems // International Research Journal. 2024. № 3(141). DOI: https://doi.org/10.23670/IRJ.2024.141.31;
9. Moreau D., Wiebels K., Boettiger C. Containers for computational reproducibility //Nature Reviews Methods Primers. 2023. Vol. 3. № 1. P. 50
