Skip to content
Standard Cloud

The Tech Race - KISS

Today, many people believe that using more technology leads to better results. I work as a cloud engineer, designing secure, reliable, and cost-effective cloud systems with the latest technologies. In my spare time, I collaborate with researchers who need precise and simple tools that are easy for non-coding experts to use. My experience shows that new, fancy technology is not always the best solution.

Different Technology Needs

For Cloud and DevOps Teams

In my job, I work with data scientists, engineers, and developers. They want and sometimes need the newest trends like Kubernetes, managed services, and large language models. They often change their tools quickly because they follow the latest trends.

For Researchers

In my free time, I work on projects for researchers who are not concerned with the latest trends. They prefer tools that work well, have few bugs, and are easy to use and improve. For them, stability, precision, and good documentation are the most important factors. They create custom software to conduct experiments, but their primary goal and expertise lie in analyzing the results.

When Modernization Goes Wrong

I often hear, “Let’s rewrite this software to be modern.” However, many good systems are thrown away just because they are not trendy. Changing to a new tool does not always mean the system becomes better. Sometimes, more technology makes the system more complicated without real benefit. I like to remind people that Facebook’s codebase is not small, yet it runs on PHP.

A Real Case: Migration from On-Prem to Cloud

Recently, I migrated a Django-based application from an on-premises VM to the cloud. Here is what happened:

  • Old System:
    The application used PostgreSQL, Celery, and Redis. It was deployed using docker-compose. Celery and Redis were reset during each deployment since they did not store any critical data.

  • Cloud Migration: I moved the database to an Azure PostgreSQL Flexible Server, which made backup easier but needed a larger server for the same performance then on premise. I also replaced Celery and Redis with Azure Container Jobs and Azure Container Apps. I wrote a lot of Terraform code and used GitHub Actions for deployment.

  • Result:

    • The cloud system offered managed backups, but it added many layers of extra code and steps.
    • For developers, nothing much changed in the application itself. However, adding new tasks became harder because it required updating Terraform scripts.
    • The cloud system turned out to be more complicated and far more expensive than a simple VM with automated backups on an Azure Storage Account using an Ansible script triggered by a scheduled GitHub Action.
    • Users lost the ability to run the application locally in the same way it runs in production. Previously, with the help of a devcontainer, developers could run the application locally with a single command by sharing the docker-compose configuration used in production.

The Value of Simplicity

This experience reminds me that simple solutions often work best. A system running on a VM can be more reliable, easier to maintain, and cheaper than a cloud system full of extra tools. Technology should be chosen to fit the needs of the business and users, not just because it is new or trendy.

Note: Although the system is simpler, a VM generally offers a higher SLA than the combined SLAs of the PostgreSQL Flexible Server, Azure Container Jobs, and Azure Container Apps.

Conclusion

In the tech race, using the latest technology is not always the right choice. Sometimes, simple is better. When planning your next project, ask yourself:

  • What real value does a new tool bring?
  • Is the new technology really needed?
  • Are the extra costs and complexity worth it?

Keep technology as a tool to help you work better. Remember: keep it simple and keep it working.

Note: Cost is not only about money. It also includes time, effort, and the risk of failure.