What We Set Up
Say a developer fixes a bug in an online store's shopping cart. With the pipeline we build, the change goes through these steps:
- the developer pushes the change to the Git repository;
- the system automatically runs checks and tests;
- if everything passes, the change goes to a staging server where you can review it;
- once approved, the change goes live, and with the right setup, with zero downtime for visitors;
- if something goes wrong, the previous version is restored in one step.
Around that flow we deliver:
- automated checks and code style tests on every change;
- automated deployment to staging and production at the push of a button or after approval;
- Docker containers, so development, staging and production match;
- separate staging environments where you and QA can see changes before they go live;
- fast rollback to the previous version;
- passwords and access keys moved out of the code into secure storage;
- documentation, so the whole team can use the process, not one person.
We pick the tool to fit your team: GitHub Actions, GitLab CI, Jenkins or another.
Which Level of Automation Fits You
- Simple automated deployment. Code from the repository goes to the server automatically after approval. Minimal effort, and it already eliminates manual file copying.
- A pipeline with checks and a staging server. Adds automated tests and a separate environment for review. The best fit for most commercial projects.
- Full containerization. Every part of the project runs in containers, and the infrastructure is defined as code, for example with Terraform or Ansible. It offers the most flexibility but takes more time to set up and more expertise to maintain.
We start with the parts where environment differences already cause problems, and we show your team how to work with containers, because they add a layer that needs maintaining. For most projects, Kubernetes is overkill, and we won't propose it unless your system needs it.
Who Needs This, and Who Doesn't Yet
Automation pays off when several developers work on a project, updates ship often, and mistakes on the live site are costly: ecommerce stores, online services, portals and mobile apps with their own backend. It is especially useful when development is split among several contractors or your in-house team, because everyone sees what others changed and the change history settles disputes.
A small corporate website updated a few times a year doesn't need a complex pipeline. Even there, we suggest keeping the code in Git with a simple automated deployment: it is inexpensive and saves you from accidental mistakes.
What to Prepare on Your Side
- Access to the code repository, or the code itself if it isn't in Git yet.
- Access to the servers or cloud account where the project runs.
- Information about who currently develops and deploys changes.
- A list of external services and access keys the application uses.
How We Roll It Out Without Stopping Development
We review how changes reach the server today, tidy up the repository so all changes go through Git with approval rules, and run the new pipeline on staging first. It is tested on several real changes before production is connected. Development continues alongside, and your team switches over only when the pipeline is proven.
What Affects the Cost
- Project state: whether the code is in Git, whether there are tests, how hard it is to spin up the project from scratch.
- Number of components: one website or several services, background workers, a mobile app.
- Number of environments: production only, or staging and demo copies too.
- Depth of automation: simple deployment, a pipeline with checks, or full containerization with infrastructure as code.
We give an estimate after reviewing your repository and current deployment process.
Frequently asked questions
Will development have to stop during implementation? No. We set up the pipeline alongside ongoing work and switch the team over once it is tested.
Can CI/CD be added to an older project? Yes, although sometimes the code and dependencies need cleaning up first. For very outdated systems, this is usually part of gradual modernization.
Do containers require the cloud? No. Docker containers run perfectly well on a regular rented server. The cloud is useful when you need to add resources quickly; see our cloud infrastructure page.
Do we need to write automated tests right away? Not necessarily. A pipeline is useful even without them: it eliminates manual deployment and gives you fast rollback. We add tests gradually, starting with the scenarios that matter most to the business, such as checkout, payments and sign-up.
Who will maintain the pipeline after setup? Your team, using our documentation, or us as part of technical support.
Which CI/CD tool will you use, GitHub Actions or GitLab CI? We choose with you, based on where your code lives and what your team already uses.
Why Promodex
We build websites and applications ourselves, with 750+ projects launched since 2015, so we set up automation from the point of view of the team that works with it every day. An in-house team does it all, with no outsourcing. You get a deployment process the whole team understands, plus a clear description of what was set up and why.