You have a complex monolithic system that is critical to your business. You’ve read articles and would love to move it to a more modern platform using microservices and containers, but you have no idea where to start.
If that sounds like your situation, then this is the article for you. Below, I identify best practices and the areas to focus on as you evolve your monolithic application into a microservices-oriented application.
We all know that net new, greenfield development is ideal, starting with a container-based approach using cloud services. Unfortunately, that is not the day-to-day reality inside most development teams. Most development teams support multiple existing applications that have been around for a few years and need to be refactored to take advantage of modern toolsets and platforms. This is often referred to as brownfield development.
Not all application technology will fit into containers easily. It can always be made to fit, but one has to question if it is worth it. For example, you could lift and shift an entire large-scale application into containers or onto a cloud platform, but you will realize none of the benefits around flexibility or cost containment.
Document All Components Currently in Use
Our newly-updated eBook walks you through incorporating containers into your CI/CD pipeline. Download the eBook
Taking an assessment of the current state of the application and its underpinning stack may not sound like a revolutionary idea, but when done holistically, including all the network and infrastructure components, there will often be easy wins that are identified as part of this stage. Small, incremental steps are the best way to make your stakeholders and support teams more comfortable with containers without going straight for the core of the application.
Examples of infrastructure components that are container-friendly are web servers (ex: Apache HTTPD), reverse proxy and load balancers (ex: haproxy), caching components (ex: memcached), and even queue managers (ex: IBM MQ).
Say you want to go to the extreme: if the application is written in Java, could a more lightweight Java EE container be used that supports running inside Docker without having to break apart the application right away? WebLogic, JBoss (Wildfly), and WebSphere Liberty are great examples of Docker-friendly Java EE containers.
Identify Existing Application Components
Now that the “easy” wins at the infrastructure layer are running in containers, it is time to start looking inside the application to find the logical breakdown of components. For example, can the user interface be segmented out as a separate, deployable application? Can part of the UI be tied to specific backend components and deployed separately, like the billing screens with billing business logic?
There are two important notes when it comes to grouping application components to be deployed as separate artifacts:
Inside monolithic applications, there are always shared libraries that will end up being deployed multiple times in a newer microservices model. The benefit of multiple deployments is that each microservice can follow its own update schedule. Just because a common library has a new feature doesn’t mean that everyone needs it and has to upgrade immediately.
Unless there is a very obvious way to break the database apart (like multiple schemas) or it’s currently across multiple databases, just leave it be. Monolithic applications tend to cross-reference tables and build custom views that typically “belong” to one or more other components because the raw tables are readily available, and deadlines win far more than anyone would like to admit.
Upcoming Business Enhancements
Once you have gone through and made some progress, and perhaps identified application components that could be split off into separate deployable artifacts, it’s time to start making business enhancements your number one avenue to initiate the redesign of the application into smaller container-based applications which will eventually become your microservices.
If you’ve identified billing as the first area you want to split off from the main application, then go through the requested enhancements and bug fixes related to those application components. Once you have enough for a release, start working on it, and include the separation as part of the release.
As you progress through the different silos in the application, your team will become more proficient at breaking down the components and making them in their own containers.
When a monolithic application is decomposed and deployed as a series of smaller applications using containers, it is a whole new world of efficiency. Scaling each component independently based on actual load (instead of simply building for peak load), and updating a single component (without retesting and redeploying EVERYTHING) will drastically reduce the time spent in QA and getting approvals within change management. Smaller applications that serve distinct functions running on top of containers are the (much more efficient) way of the future.
Vince Power is a Solution Architect who has a focus on cloud adoption and technology implementations using open source-based technologies. He has extensive experience with core computing and networking (IaaS), identity and access management (IAM), application platforms (PaaS), and continuous delivery.
Join us for free online training courses, hosted monthly by a Rancher technical expert. We provide a great hands-on overview for new users setting up a Rancher deployment, and answer any and all questions you have about Rancher and how to integrate it into your DevOps processes!