If the Service template and Microservice chassis are the solution, what is the problem?
Last week, I had a great interview with Sven Johann for the CaSe: Conversations about Software Engineering podcast. The topic of the interview was the Service template and Microservice chassis patterns. As part of the preparation for the interview, I reviewed the two patterns on Microservices.IO and I was surprised to see that the both patterns were missing an essential element: the problem statement. A situation that reminds me of the number 42 from The Hitchhiker’s Guide to the Galaxy.
About the Service template and Microservice chassis patterns
Here are the thumbnail definitions of the two patterns.
The Service template pattern thumbnail:
Quickly create services by cloning a service template, which is a runnable, template code base that implements the logic that is common to each service.
The Microservice chassis pattern thumbnail:
Improve maintainability of services cloned from a service template by extracting the logic from the template into a microservice chassis framework.
This diagram shows the structure that results from applying these patterns:
The key elements are as follows:
- Microservice chassis - a framework implements the bulk of the logic common to each service
- Service Template - a template codebase that uses the Microservice chassis
- Service - one or more services that are cloned from the Service Template
What problem do these patterns solve?
Both pattern descriptions contain all of the essential elements (Context, Forces, Solution, Consequences, Related pattern) of a pattern except for the problem statement. So what is a concise way of stating the problem? My first thought was a suitable problem statement would be “How can a team quickly create a new service?”. While this is simple and concise, it’s somewhat vague. What does it mean to “create a service”?
If you read the context that’s shared by the two patterns, it describes the challenges with creating and maintaining a service’s code base. In particular, how to setup up and maintain the the service’s build logic and the code that implements common cross cutting concerns. But why is that a worthwhile problem to solve?
It’s important to reduce the time spent setting up this logic, because it enables the team to quickly start working on the service’s business logic - its reason for existing. What’s more, it’s important to solve this problem in a way that simplifies the future maintenance of many services. Consequently, I thought the following is an accurate problem statement:
How can a team quickly create and setup a maintainable code base for a production-ready service so they can start developing its business logic?