Technology leaders need to pay attention to DevOps. Yes, it’s a funny little name. Wikipedia states that DevOps is a compound of “development” and “operations” before explaining it as “a culture, movement or practice that emphasizes the collaboration and communication of both software developers and other information-technology (IT) professionals while automating the process of software delivery and infrastructure changes.”
Technology professionals know that identifying, tracking and resolving bugs costs money. If you are the one writing the software (and sooner or later, everyone will), the bugs are on your dime. Good testing practices can help minimize bugs and costs. However, sometimes bugs result from deployment practices. Indeed, the best technology operations focus on standardized, automated testing and release management practices. By DevOps best practices, software teams treat software deliverables the way a manufacturing company treats finished goods – ruthlessly eliminating deviations with automation.
If you have tried and failed to create innovative solutions within your company by writing software, there could be several reasons why that happened. If you think you got the requirements right, and think the architecture was right, and your software developers understand the technology, then examine the process of delivering the software to the users.
Delivering Software Cost Effectively
The concept behind DevOps has been known as Continuous Integration (CI), Application Lifecycle Management (ALM) and by other names. Often, IT departments found ALM complex, or did not have the knowledge required to design a pipeline for software development. But, the tools have continued to evolve, and the processes have simplified. Today, Cloud vendors deliver DevOps services to technology professionals which are very hard to dismiss. Among the very best is Microsoft’s Azure platform. Microsoft Azure provides many tools for standardizing, testing and delivering high quality software.
Here are my four favorites:
Azure Resource Management (ARM) templates
Azure Resource Management templates are JSON documents which can be used to describe a complete set of Azure services. These documents can be saved and managed by IT operations personnel. This highlights a key cloud computing value proposition: the cloud offers technology as a “standard service” and each service can be encapsulated to be brought up and down as needed.
ARM templates can describe Infrastructure-as-a-Service offerings (i.e. Virtual Machines, Networks and Storage). This enables Dev / Test Labs to be designed, templated, deployed and undeployed as needed. Technology teams which must plan for an upgrade by providing a test environment no longer need to buy infrastructure to support a virtual environment. Instead, they can define the environment as an ARM. Azure allows you to build the environment once, extract the ARM template for later use, and then destroy the resources.
ARM templates can describe Platform-as-a-Service offerings (i.e. Websites, Services, Databases). This enables the exact same concept, with even better results. In the end, you don’t even have any servers to manage or patch: the underlying infrastructure is standardized. This brings me to Deployment Slots.
A common best practice in delivering software is to have at least one Quality Assurance (QA) environment. This shadow environment should replicate production as closely as possible. However. in the PaaS world, we don’t have control of the underlying infrastructure – that’s great, it’s standardized and we want to keep it that way. But we don’t want to abandon the practice of performing final testing before deploying to production.
With deployment slots, we get the ability to create a number of “environments” for our applications and services, then switch them back and forth as needed. Let’s say you have a new software release which you want to ensure passes some tests before releasing to the user community. Simply create a slot called “Staging” for deployment, perform your tests, then switch to production.
Uh oh – we missed something. We’re human after all. Users are reporting bugs and they liked it better the way we had it. Switch it back – no harm no foul.
There are some important things to consider before adding Deployment Slots to your DevOps pipeline. For example, if your application relies upon a database of some kind, you may need to provision staging copy for your tests. You also need to be aware that Connection Strings are one of the configuration values which can switch with the slot, unless configured to do otherwise.
Deploy to Azure
I was recently treated to some excellent material on the Cortana Analytics Suite of products. Paying close attention (as I sometimes do), I noticed that the lab environment was prepared for me as an ARM template. I was directed to GitHub (an online public software repository) and told to push the button marked “Deploy to Azure”. When I did, I was brought to http://deploy.azure.com – and the URL included a reference to the GitHub location, or repository, which I had just visited. The author of the software had placed an ARM template describing the entire lab environment, and included a few parameters so that I could fill in the information from my Azure subscription. 20 minutes later, I had Machine Learning, Hadoop/Spark, Data Factory and Power BI resources at my fingertips. Later in the day, we did deployed again, this time deploying a simple Web app which consumed Advanced Analytics services. When I was finished, I simply deleted the resources – the entire day cost me less than $20 of Azure consumption costs. Deploying an app has never been easier.
Azure Container Services
No discussion of DevOps would be complete without mentioning Docker. Docker is a platform gaining popularity among developers and IT operations for its consistency with Virtual Machines and lower overhead. Essentially, Docker runs as a subsystem which hosts containers. A container is similar in functionality to ARM.
DevOps Tools on Azure
Linux or Windows, Open Source or Closed, Infrastructure or Platform, TFS or GitHub. None of that matters anymore. No more excuses – Microsoft Azure provides outstanding DevOps tooling for Modern Application Development. If you have not deployed your first application to Azure, let’s talk. We can get you optimized quickly
About Brian: Brian Berry leads the Microsoft Business Intelligence and Data Analytics practice at BlumShapiro. He has over 15 years of experience with information technology (IT), software design and consulting. Brian specializes in identifying business intelligence (BI) and data management solutions for upper mid-market manufacturing, distribution and retail firms in New England. He focuses on technologies which drive value in analytics: data integration, self-service BI, cloud computing and predictive analytics.