Traditionally, software development teams are split into two dedicated functions — Software Engineers and Operations/Information Technology (Ops/IT) Specialists. Software engineers were responsible for writing code to solve a customer problem. Once the code was ready, an Ops/IT professional would deploy it to production servers and monitor it to ensure everything was running smoothly. It’s easy to forget that there used to be physical networking equipment and servers for every project not so long ago. At the time, software groups were not equipped to handle the technical challenges of working with these physical devices.
These roles are now blurring with the adoption of cloud computing services, giving birth to a new discipline known as Developer Operations Engineering or DevOps. Using software as their main tool, DevOps engineers work on internal development problems. Similar to how a software engineer optimizes an algorithm, a DevOps engineer’s goal is to optimize the software engineering process. In other words, dedicated developers use software to solve customer problems and DevOps engineers use software to solve their team’s software engineering problems. Since the tools used by DevOps are usually code-based, it’s becoming a critical skill for any software developer.
Operations as development
When defining a new software system, engineers must decide what platform they’ll use to deploy. Different applications require different resources. For example, an image processing application might require specialized equipment like a graphics processing unit (GPU). These decisions have historically been made by system administrators and set up by hand. More recently, DevOps engineers have been responsible for these decisions and written template files that define the architecture. Rules can be written to scale system capabilities to address how systems should react to high volumes of user requests. Writing these files makes the system clearly defined, extensible, and reproducible. Multiple programming languages like Python, Bash, webpack, and gulp are used by DevOps teams to achieve these goals.
In the Cloud Developer Nanodegree program, we cover a variety of technologies from databases and filestores to various computation platforms like Docker, AWS EC2, and AWS Lambda (Serverless). An understanding of these platforms is a requirement for becoming a DevOps engineer and we believe these skills are fundamental to developing on the cloud.
More development and fewer operations
DevOps engineers are usually obsessed with automating the medial tasks that slow down software development. One frequent automation is a code deployment pipeline. Known as continuous integration/continuous deployment (CI/CD), these pipelines monitor version control system changes. The system checks the changes for accuracy using automated testing called “unit tests.” These tests are run in a dedicated test environment where the broken code can’t affect users.
The release process is the final step. The “CD” part of the system deploys the tested code to production servers—which provide the software product to real users. By automating this task, engineering teams can deploy multiple times a day or quickly respond to critical bug requests. Travis CI and CircleCI are some popular tools used to automate deployment.
Again, we believe DevOps skills are fundamental for any cloud developer. In my course for full-stack engineering on Amazon Web Services (AWS) in Udacity’s Cloud Developer Nanodegree program, I lay the framework for the skills you’ll need as a cloud developer to begin understanding how to practice DevOps for your cloud deployments. In the course, we build an imageboard with image style filters using modern technologies like TypeScript and Node to build the server. We cover how to write scripts to build the code for production and then use AWS Elastic Beanstalk to configure infrastructure and deploy.
Any Dev Can DevOps
Learning necessary DevOps is a logical professional development leap for any developer:
- Frontend developers can ship and test features faster
- Fullstack engineers can deliver updates without breakage concerns
- Machine learning researchers can deliver new, more accurate models and optimize infrastructure
Although some teams have dedicated DevOps groups, many rely on their top software engineers for this function. This shared responsibility promotes a DevOps culture across all team members. Usually, there are a few high caliber engineers who begin doing less Dev and more DevOps and assume the title DevOps Engineer. DevOps allows you to help your team reduce costs, improve collaboration, minimize downtime, and decrease the time to ship new features. We’ve taken care to design the Cloud Developer curriculum to lay out a path towards DevOps in your engineering practice.
