Azure DevOps – What It Offers
Who should actually use Azure DevOps? In most cases it is companies already working with the Azure platform or other Microsoft services, since for them it simply extends the tool stack they have in a natural way.
The IT industry has plenty of tools that help with the processes of an application’s life cycle. They let us automate the tasks we keep repeating and stay in control of the entire process. Broadly, they fall into two groups: tools covering only individual pieces of the process, and platforms capable of managing it from start to finish. If you are looking for a toolset that will make delivering IT solutions faster, the Microsoft platform Azure DevOps may be worth a look.
What is Azure DevOps?
Azure DevOps is a SaaS platform on which you can run every one of your DevOps processes. DevOps itself is a mix of ideas, tools and practices for building and releasing applications automatically, quickly and with high quality. Azure DevOps was built to back each DevOps process and its building blocks so that the application life cycle runs as efficiently as it can. There are 5 core services, each supporting a different stage of that life cycle. Let’s walk through the process and see how these services can make our job easier and quicker.
Azure Boards
Every application begins with planning – before anyone starts working, we have to know what has to be done, what our goals are and what business value the planned product will bring. That is exactly where Azure Boards proves useful.
image source
Here you set up the processes and the workflow the team will follow when building the application. You give every team member a role and divide the work into tasks. Azure DevOps projects come with several workflows – Agile, Basic, CMMI and Scrum – so you can pick the one that fits your team or organization best.
An organization may contain many projects, and a single project may contain many teams. Azure Boards handles several projects and several teams inside one organization. Thanks to its built-in integration with the other Azure DevOps components, it turns out to be a very handy tool for communication between everyone on the project, such as developers, testers, analysts or product owners. With Azure Boards you get a clear, transparent view of where a feature stands while it is still being built.
At first glance this is a fairly typical workflow-support tool, yet Azure Boards has some notable features too. One of them is the option to enable the DevSecOps methodology, offered together with Github. Among other things, it covers protecting application code running in production and makes it possible to scan both the code and third-party components for vulnerabilities.
See the Microsoft documentation for Azure Boards.
Azure Repos
image_source
At the heart of the application lifecycle is the code that developers write. To host it, Azure offers Azure Repos. It works with a variety of workflows, git-flow among them, all of which are meant to produce code of the highest possible quality. As with most present-day repository tools, Azure Repos helps developers work together: besides storing the code, it lets you link commits and pull requests to particular requirements or tasks in Azure Boards and to tests in Azure Tests.
See the Microsoft documentation for Azure Repos.
Azure Pipelines
image_source
When a feature is finished and its pull request has been approved and merged into the main branch, the code has to be released. For that, it must be tested and packaged into an artifact, which can then be deployed to the target environment. This is what makes the feature available to end users. Azure provides a way to automate all of this, called Azure Pipelines – the place where the CICD process gets automated. A good starting point is Continuous Integration. Pipelines are defined in YAML files, so we can keep them in the code repository, version them and treat them as part of our codebase.
The basic building blocks of Azure pipelines are steps. A step can handle many kinds of work, for example running tests, packaging an application, building images or pushing images to repositories. Steps can be written as scripts or run as tasks, which are ready-made commands created to carry out particular actions in the pipeline. Azure Pipelines ships with a broad set of such tasks for various scenarios, including building, testing and deploying applications. You can pick and configure them easily through a graphical user interface (GUI) in the Azure DevOps portal, which makes creating and maintaining pipelines simpler. As a result, users can assemble complex pipelines with very little code or scripting, and the development process becomes more efficient and smoother. If you like, you can have them converted to yaml format automatically, then tweak them or add configuration, and in the end push the result to the repo.
If you work with several jobs, they have to be declared under job attributes. A job can contain many steps, and all of them can run within that job on an agent. This makes it possible to run several processes in parallel and boost the pipeline’s overall speed and efficiency. By declaring a number of jobs and steps in the YAML file, we can build a richer, more customizable pipeline tailored to what our project needs.
The agent is the machine that runs the pipeline’s tasks or steps. The pool attribute lets us specify which kind of agent should run the steps (i.e. Linux, MacOS, Windows), as well as particular versions of those operating systems.
A typical reason to use several jobs in Azure Pipelines is a group of steps we would like to run in parallel so that the whole build finishes sooner. Splitting the work into separate jobs lets us spread the load and run the steps at the same time, which can make the pipeline perform better.
See the Microsoft documentation for Azure Pipelines.
Azure Artifacts
image_source
What artifacts the team produces depends on the programming language of the application. To store them, there is yet another feature, Azure Artifacts. If your artifacts are of one of these types:
- NuGet
- npm
- Maven
- Python
- Universal Packages
you can keep them in Azure Artifacts once your pipeline has produced them.
Quite often, though, a build pipeline ends up producing a Docker image. In that case a dedicated repository is required. The simplest choice is Azure Container Registry, but if you need an external repository such as Docker Hub, you will have to set up a connection to it from Azure DevOps.
See the Microsoft documentation for Azure Artifacts.
Azure Test Plans
image_source
Testing is a crucial element of releasing an application. Azure DevOps includes a separate component built for this – Azure Test Plans. Application code has to be tested thoroughly at every stage of development, from smoke tests in CI all the way to acceptance tests on pre-production. Test Plans handles both manual and automated tests. It gives you one central view of every test case that has to pass before production deployment gets the go-ahead.
This is where you create manual test cases and test plans. Once a new feature is ready, a tester can follow the steps and check the application.
It can also hold automated tests that run as part of the CICD process. Test reports produced by the pipelines can be displayed in Azure Test Plans as well.
A major benefit is the integration with Azure Boards, which lets you launch your tests from there, for instance from the Kanban board.
See the Microsoft documentation for Azure Test Plans.
Architecture
When a pipeline runs, its configured tasks need some place to execute. I mentioned before that they can run, for example, on various operating systems. So where do these tasks actually run? Azure DevOps is SaaS, which means all of its components are delivered as a service, and every configuration, pipeline and repository lives in that service. For running pipelines, Azure DevOps offers two choices. You can use machines (agents) that Microsoft manages, or self-hosted agents that you configure to connect to the Azure DevOps service. The latter may be the way to go if you need full control over your agents. Both kinds of agents can also be used side by side.
Integrations
Out of the box, Azure DevOps covers nearly the entire application development cycle. But what if your process already relies on other tools? Say your code already lives in a different repo, and some time after the project kicked off you decided to move to Azure DevOps. Do you then have to migrate everything you have already started to the platform? Not really. Azure DevOps can integrate with other services. You can connect its components to, for example, Github or Docker Hub, or deploy to other cloud platforms such as GCP or AWS. This is what the service connection feature is for.
In Azure DevOps, the service connection feature makes connecting to external services easier. It takes care of the credentials required for those services, which Azure creates automatically on its side, and it makes connections more secure because Azure DevOps manages the connection secrets. This reduces the chance of unintentionally leaking sensitive data or misconfiguring connection settings. Service connections work with many kinds of external services, such as cloud providers, code repositories, build and release systems, and monitoring and analytics tools.
You can also set permissions for each connection individually within your organization and project.
See the Microsoft documentation for Azure Pipelines Service Connection.
Azure DevOps comparison
Many tools can support application development across a number of popular workflows. Answering with a plain “yes” or “no” why someone should learn and adopt Azure DevOps is hard, because it depends on many factors and circumstances. Still, a short comparison with competing products can show what makes it different. The market offers CICD tools such as Jenkins, Github Actions, Argo CD and Bamboo. The key difference is that these are purely CICD tools. They excel at automating tasks and running them at intervals. DevOps, however, is a set of activities stretching across the entire application life cycle. Azure DevOps gives you an all-in-one solution combining project management, source control, continuous integration and delivery, testing and deployment. On top of that, it integrates natively with other Azure services, which can make building, testing and deploying applications on Azure easier. You manage every piece of it from a single place, which is extremely convenient.
There is, however, one more contender – AWS DevOps. It is built into AWS and tied directly to your billing account, so you need an AWS account to use it. Azure DevOps, by contrast, is a standalone product that you can manage independently of your Azure account. In other words, you can use Azure DevOps without using Azure, and the other way round. That said, AWS and Azure DevOps provide comparable features and capabilities, such as continuous integration and delivery, testing and deployment, and each integrates natively with its own cloud platform. For instance, Azure DevOps hosted runners run on Azure, and code kept in Azure DevOps repositories is hosted on Azure too.
Azure Active Directory can be integrated with Azure DevOps. If, however, you need to deploy from Azure DevOps to, say, AWS EKS, you must connect to it the same way you would to any other Cloud platform when deploying to multiple platforms.
Summary
With everything described above in mind, let’s try to answer the question: who is Azure DevOps meant for?
In most cases, companies already using the Azure platform or other Microsoft services will gravitate towards Azure DevOps, because it extends their existing tool stack in a natural way. Likewise, AWS users will most likely go for AWS DevOps. For other projects whose processes are already set up, for instance with open source solutions, moving everything into a single integrated service is unlikely to pay off. In the end, picking between AWS DevOps and Azure DevOps may come down to things like how well you know each platform, what your team specifically needs, and how much each platform costs.