Hello, we are using Keycloak in our systems and are very happy with how powerful is this tool and how rich is the functionality. However, when you use such a tool in the enterprise there are some customizations required, also, there are some known bugs in the project. We are trying to contribute to the project to make it even better but sometimes it might be difficult.

Keycloak has a very rich functionality on many different levels, but on the other hand, it makes the project huge and monolithic in some sense. In one project there are APIs, themes, protocols, infrastructure, tests, etc

Proposal

Split the Keycloak project into smaller ones and aggregate them in the main project (this one)

Benefits

I see many benefits from this, such as:

• Easier to contribute. It'll be easier to contribute, projects will be smaller, and it'll be easier to grasp the purpose and ideas of the single project than the whole Keycloak and how it works inside in general. Tests and code checks will run faster because they will be for separate smaller projects. I think that the easier contribution -> the more contributors -> the faster and stronger the project is getting and more features and bugfixes it obtaines.
• Re-usage by the community. Separate projects can be reused somewhere if needed not only in the Keycloak which can be a good help for the community. For example, Keycloak is OIDC compliant, it has all the grant flows implemented as well as strong and secure enough to be used in the big enterprises. But what if somebody wants to create their own JAVA-based OIDC-compliant service? They could reuse or build it based on what is currently in the Keycloak knowing that it's a solution forged by time. If OIDC implementation of the Keycloak would be a separate project that would be possible!
• Better modules and code independence. I believe that having code split into separate projects enforces looser coupling and better cohesion of the code. This makes it easier to test, easier to understand and maintain. This makes it easier to plug in custom implementations, for example, adding the support for NoSQL databases, and have Admin panel in Vue.js instead of Angular, etc
• More options for monetization and enterprise solutions. Since increasing the modularity can make it easier to have new implementations for some parts, those parts can be developed by the Keycloak team and sold in a marketplace (for example, Red Hat Marketplace). For example, if there is a need to use, let's say, Redis instead of Infinispan, there can be a separate module developed and distributed under a commercial license. Can be an additional income for the Red Hat ;)

Candidates

Here are some candidates that can become separate projects:

• Themes: Admin, account, email. Maybe even login theme.
• OIDC protocol
• SAML protocol
• Admin REST API
• User API
• Testing infrastructure. For example, to make it easier to test Keycloak with custom modules. Currently, we have Keycloak with many custom modules / providers / etc. Instead of creating our own test framework, we could reuse the experience of the Keycloak team and use a testing module with already built-in Keycloak-specific helpers.
• Data access layer (DB / Cache)
• Infrastructure / framework. So if there will be a new idea on the horizon for a new framework (as it was for Wildfly -> Quarkus) it'll be easier to swap.
• Event system
• Secret manager. From what I know there is no central secret manager at the moment and it would be great to have one, for example, to keep private keys and client secrets in the secure place

Proposed changes can be started small, let's say, from one module, to see how it goes, something simple or having very few dependencies.

Challenges to overcome

• requires architecture change
• more projects might be harder to maintain from the management point of view

Let me know if some of the above make sense and might be a way to go for the future. Also, if you disagree, or think that it's not the right way to go, I would be glad to hear you out.

In any case, looking forward to feedback and am willing to contribute!