As Drupal has evolved, it has become more than just a CMS. It is now a fully fledged Web Development Platform, enabling not just sophisticated content management and digital marketing capabilities but also any number of use cases involving data modelling and integration with an endless variety of applications and services. In fact, if you need to build something which responds to an HTTP request, then you can pretty much find a way to do it in Drupal.
“Just because you can, doesn’t mean you should.”
However, the old adage is true. Just because you can use use a sledgehammer to crack a nut, that doesn’t mean you’re going to get the optimal nut-consumption-experience at the end of it.
Drupal’s flexibility can lead to a number of different integration approaches, all of which will “work”, but some will give better experiences than others.
On the well trodden development path of Drupal 8, giant steps have been taken in making the best of what is outside of the Drupal community and “getting off the island”, and exciting things are happening in making Drupal less of a sledgehammer, and more of a finely tuned nutcracker capable of cracking a variety of different nuts with ease.
In this post, I want to explore ways in which Drupal can create complex systems, and some general patterns for doing so. You’ll see a general progression in line with that of the Drupal community in general. We’ll go from doing everything in Drupal, to making the most of external services. No option is more “right” than others, but considering all the options can help make sure you pick the approach that is right for you and your use case.
Build it in Drupal
One option, and probably the first that occurs to many developers, is to implement business logic, data structures and administration of a new applications or services using Drupal and its APIs. After all, Entity API and the schema system give us the ability to model custom objects and store them in the Drupal database; Views gives us the means to retrieve that data and display it in a myriad of ways. Modules like Rules; Features and CTools provide extensive options for implementing specific business rules to model your domain specific data and application needs.
This is all well and good, and uses the strengths of Drupal core and the wide range of community contributed modules to enable the construction of complex sites with limited amounts of coding required, and little need to look outside Drupal. The downside can come when you need to scale the solution. Depending on how the functionality has been implemented you could run into performance problems caused by large numbers of modules, sub-optimal queries, or simply the amount of traffic heading to your database - which despite caching strategies, tuning and clustering is always likely to end up being the performance bottleneck of your Drupal site.
It also means your implementation is tightly coupled to Drupal - and worse, most probably the specific version of Drupal you’ve implemented. With Drupal 8 imminent this means you’re most likely increasing the amount of re-work required when you come to upgrade or migrate between versions.
It’s all PHP
Drupal sites can benefit hugely from being part of the larger PHP ecosystem. With Drush make, the Libraries API, Composer Manager, and others providing the means of pulling external, non-Drupal PHP libraries into a Drupal site, there are huge opportunities for building complexity in your Drupal solution without tying yourself to specific Drupal versions, or even to Drupal at all. This could become particularly valuable as we enter the transition period between Drupal 7 and 8.
In this scenario, custom business logic can be provided in a framework agnostic PHP library and a Naked Module approach can be used to provide the glue between that library and Drupal - utilising Composer to download and install dependencies.
This approach is becoming more and more widespread in the Drupal community with Commerce Guys (among others) taking a libraries first approach to many components of Commerce 2.x which will have generic application outside of Drupal Commerce.
The major advantage of building framework agnostic libraries is that if you ever come to re-implement something in another framework, or a new version of Drupal, the effort of migrating should be much lower.
Building on the previous two patterns, one of Drupal’s great strengths is how easy it is to integrate with other platforms and technologies. This gives us great opportunity to implement functionality in the most appropriate technology and then simply connect to it via web services or other means.
This can be particularly useful when integrating with “internal” services - services that you don’t intend to expose to the general public (but may still be external in the sense of being SaaS platforms or other partners in a multi-supplier ecosystem). It is also a useful way to start using Drupal as a new part of your ecosystem, consuming existing services and presenting them through Drupal to minimise the amount of architectural change taking place at one time.
Building a solution in this componentised and integrated manner gives several advantages:
- Separation of concerns - the development, deployment and management of the service can be run by a completely separate team working in a different bounded context. It also ensures logic is nicely encapsulated and can be changed without requiring multiple front-end changes.
- Horizontal scalability - implementing services in alternate technologies lets us pick the most appropriate for scalability and resilience.
- Reduce complex computation taking place in the web tier and let Drupal focus on delivering top quality web experience to users. For example, rather than having Drupal publish and transform data to an external platform, push the raw data into a queue which can be consumed by “non-Drupal” processes to do the transform and send.
- Enable re-use of business logic outside of the web tier, on other platforms or with alternative front ends.
Headless Drupal is a phrase that has gained a lot of momentum in the Drupal community - the basic concept being that Drupal purely responds with RESTful endpoints, and completely independant front-end code using frameworks such as Angular.js is used to render the data and completely separate content from presentation.
Personally, I prefer to think of a “nearly headless” approach - where Drupal is still responsible for the initial instantiation of the page, and a framework like Angular is used to control the dynamic portion of the page. This lets Drupal manage the things it’s good at, like menus, page layout and content management, whilst the “app” part is dropped into the page as another re-usable component and only takes over a part of the page.
For an example use case, you may have business requirements to provide data from a service which is also provided as an API for consumption by external parties or mobile apps. Rather than building this service in Drupal, which while possible may not provide optimal performance and management opportunities, this could be implemented as a standalone service which is called by Drupal as just another consumer of the API.
From an Angular.js (or insert frontend framework of choice) app, you would then talk directly to the API, rendering the responses dynamically on the front end, but still use Drupal to build everything and render the remaining elements of the page.
As we’ve seen, Drupal is an incredibly powerful solution, providing the capability for highly-consolidated architectures encapsulated in a single tool, a perfect enabler for projects with low resources and rapid development timescales. It’s also able to take its place as a mature part of an enterprise architecture, with integration capabilities and rich programming APIs able to make it the hub of a Microservices or Service Oriented Architecture.
Each pattern has pros and cons, and what is “right” will vary from project to project. What is certain though, is that Drupal’s true strength is in its ability to play well with others and do so to deliver first class digital experiences.
New features in Drupal 8 will only continue to make this the case, with more tools in core to provide the ability to build rich applications, RESTful APIs for entities out of the box allowing consumption of that data on other platforms (or in a headless front-end), improved HTTP request handling with Guzzle improving options for consuming services outside of Drupal, and much more.