Switching to a Container-based Application Architecture? Not So Fast
November 03, 2021

Jayne Groll
DevOps Institute

Container technology, such as Docker and CoreOS, have become an increasingly popular step in DevOps journeys. Containers are lightweight and portable, making them easier on organizations' resources. Containers can benefit cloud environments in a variety of ways, but that does not mean they are perfect for every situation. Software teams must manage a container's life cycle, including provisioning, deployment, scaling (up and down), networking, load balancing and more. The maintenance and implementation of container technology can be tedious and, therefore, there are situations where containers simply are not a good idea.

I asked top industry experts — DevOps Institute Ambassadors— to share their thoughts on when you may not want to use containers. Here's what landed at the top of the list:

Anshul Lalit

head of technology and transformation, Kongsberg Digital

"Containers are perfect for deploying an application with many dependencies that the end-user does not need to touch. However, containers might not be the best option if the application is very lightweight, and it's not essential for users to quickly update the application with each new version. For example, I would recommend for a monolith lift-and-shift to use a simple VM to avoid disruption.

In addition, it is vital to understand the limitations and differences between using a container and using another technology stack. Some use cases where you may avoid containers are:

If you need to use a special driver, capability or incompatible software with the standard container kernel.

If you need to change the kernel configuration, use a custom kernel, add other software or run commands that interfere with a container's standard configuration.

If you need to work with files that are incompatible with a container. For example, if you need to access a file system or a database not available in a container.

Also, as a general guideline, I highly recommend The Twelve-Factor App methodology for building modern software-as-a-service (SaaS) apps. It just makes your product life easier, predictable and covers all bases."

Supratip Banerjee

solutions architect, Principal Global Services

"If security is a priority, maybe Docker is not the best solution.

The most significant security benefit of Docker is that it divides the software into smaller pieces. If the security of one component is jeopardized, others are unaffected. While separated processes in containers offer greater security, all containers have access to the same host operating system. You face the risk of operating Docker containers with insufficient isolation. Any malicious malware has the ability to gain access to your computer's memory.

It is common practice to operate a large number of containers in a single environment. Unless you limit the resource container capabilities, this is how you make your app vulnerable to resource abuse attacks. Each container should handle one specific area of concern for optimal efficiency and isolation.

Another issue is that Docker's default setup does not namespace users. Namespaces allow software resources to use other resources only if they are in the same namespace."

Parveen Arora

co-founder and director, VVnT SeQuor

"If you're doing a project(s) related to "lift and shift" of your applications, you may be better off with a simple VM deployment to experience the least amount of disruption. However, if you're creating a new application from scratch, you're probably better off starting with containers."

Vishnu Vasudevan

head of product engineering and development, Opsera

"If you're running a monolithic application, people assume that switching from monolithic to container-based microservices is like flipping a switch, but that's not the case. Understanding the application needs to happen first and deciding to make it containerized is only viable if your team has the right skills for converting a monolithic application into a containerized application. If you do not have the skills within the application team running the monolith, they are likely going to fail. Whoever is building the application needs to know best practices for container orchestration. They need to analyze and recommend whether or not this application fits in the wheelhouse of containerization. Do not try to move everything to containers because it's the popular thing to do or your competitors are doing it. It has to be solely based on the merit of the application, not based on the merit of the person pushing to move to containers."

Join DevOps Institute for SKILup Day: Container Orchestration, for how-to sessions to expand your container knowledge.

Jayne Groll is CEO of DevOps Institute
Share this

Industry News

December 19, 2024

Check Point® Software Technologies Ltd. has been recognized as a Leader in the 2024 Gartner® Magic Quadrant™ for Email Security Platforms (ESP).

December 19, 2024

Progress announced its partnership with the American Institute of CPAs (AICPA), the world’s largest member association representing the CPA profession.

December 18, 2024

Kurrent announced $12 million in funding, its rebrand from Event Store and the official launch of Kurrent Enterprise Edition, now commercially available.

December 18, 2024

Blitzy announced the launch of the Blitzy Platform, a category-defining agentic platform that accelerates software development for enterprises by autonomously batch building up to 80% of software applications.

December 17, 2024

Sonata Software launched IntellQA, a Harmoni.AI powered testing automation and acceleration platform designed to transform software delivery for global enterprises.

December 17, 2024

Sonar signed a definitive agreement to acquire Tidelift, a provider of software supply chain security solutions that help organizations manage the risk of open source software.

December 17, 2024

Kindo formally launched its channel partner program.

December 16, 2024

Red Hat announced the latest release of Red Hat Enterprise Linux AI (RHEL AI), Red Hat’s foundation model platform for more seamlessly developing, testing and running generative artificial intelligence (gen AI) models for enterprise applications.

December 16, 2024

Fastly announced the general availability of Fastly AI Accelerator.

December 12, 2024

Amazon Web Services (AWS) announced the launch and general availability of Amazon Q Developer plugins for Datadog and Wiz in the AWS Management Console.

December 12, 2024

vFunction released new capabilities that solve a major microservices headache for development teams – keeping documentation current as systems evolve – and make it simpler to manage and remediate tech debt.

December 11, 2024

CyberArk announced the launch of FuzzyAI, an open-source framework that helps organizations identify and address AI model vulnerabilities, like guardrail bypassing and harmful output generation, in cloud-hosted and in-house AI models.

December 11, 2024

Grid Dynamics announced the launch of its developer portal.

December 10, 2024

LTIMindtree announced a strategic partnership with GitHub.