Getting Started with Infrastructure as Code. Understanding IaC Core Concepts. When deploying a software application, we'll generally need some server and networking infrastructure configured and available to deliver our application to our users for us. Managing all this infrastructure introduces many additional points from outside of our source code for potential failure or compromise into our application. This can often make problems difficult to troubleshoot or even detect.

Deploying Infrastructure as Code resolves all of these problems with a modern DevOps approach. Employing DevOps principles for server infrastructure. The DevOps model, as the name might imply, combines the development and operations roles together. We can speed up development and reduce errors when the development team is also involved with managing the environment their code will operate on.

Infrastructure as Code inherently supports the DevOps model by making infrastructure configuration and management part of our source code. Reducing human error and saving time with automation. This DevOps model sounds like a great idea, but isn't this essentially requiring developers to do two jobs now? How do we manage this extra workload? Infrastructure as Code brings operations into the developer's comfort zone, writing configuration files and scripts to define the operating environment for our application just becomes another part of our source code. By deploying infrastructure from code like this, we can quickly and reliably automate server setup processes that were previously time-consuming manual operations prone to human error.

With Infrastructure as Code, one person can fulfill both development and operations roles more quickly and efficiently than was possible with more traditional models, where these roles often needed to be split into separate job positions in the past because of the time each required on their own.

Working with immutable server deployments. There is a very particular difference when working with Infrastructure as Code compared to more traditional server administration practices. We never want to actually log into the server and make any changes or updates to our system configuration after the server is deployed. All server configuration should only be defined in one place within our source code. Whenever our configuration code changes, we'll deploy a new server with that new configuration to replace the previous server. An analogy you can use here is to consider your servers like livestock rather than pets. With Infrastructure as Code, we are no longer pruning and caring for a single server instance for the life of our application. Instead, servers can be as short-lived as versions of our application code, and replaced every time we make a configuration change. This practice ensures that our application will run consistently whenever we need to deploy additional servers. We never have to worry about missing any custom configuration or update that was manually performed on a pet server.

Immutable servers are also faster to deploy than a server built with a configuration management tool like Puppet, Chef, or Ansible, which we'll discuss in more depth later. This is because these configuration management tools have to wait until a new server is deployed, then connect to it to perform any installation tasks or execute any scripts needed to bring the server up to the desired state. All of these updates are instead already baked into an immutable server image, so the server is ready to go the moment it's deployed without the overhead of an additional tool required to get it there.

Securing and monitoring deployments. While Infrastructure as Code solves many operational problems for us, it has a few security challenges of its own to be aware of. Most often it's the developer who gets tasked with taking on operations, rather than operations taking on development. Infrastructure as Code is logically assumed to be the responsibility of the coders. Developers however are not inherently well versed in matters of server and network security, and are often unaware of the best practices they need to be following in their Infrastructure as Code configurations.

Since IaC blurs the line between development and operations, this can also have the effect of blending passwords and other secret data into our application code during development. We need to be careful and make sure that credentials or environment variables aren't stored in plain text files along with the rest of the IaC configuration anywhere in our source code. Since we have all of our configuration in just one place now, if this single source is compromised, then our entire application becomes compromised. This is also why it's important to define explicit accounts with minimal permissions for use during development and testing, and not just use our admin credentials everywhere because it's more convenient.

Also, since a new server is deployed with every revision of our code, it's important to regularly audit our builds and make sure that old servers are properly spun down and deleted. The last thing you want is to rack up costs from a bunch of test servers that were used for ten minutes then forgotten about but left running. This is an unfortunately easy mistake to make when first starting out with Infrastructure as Code.