Posts tagged with 1.2.x

One of the less enjoyable aspects of apiman is the manual addition of an API that you wish to manage. And if you have a bunch of APIs you want to manage, you can either use the apiman REST interface to script the creation of them, or else you’re stuck manually entering them into the UI.

However, if you take advantage of the new API Catalog feature, things might get a lot easier!

One of the strongest features of apiman, in general, is its excellent extensibility. Not only is it easy to add new policies, for example, but many of its core components are also pluggable. This includes, for example, the registry used by the API Gateway to store configuration information published to it by the manager. This blog post will detail a new JDBC based implementation of that registry, explaining how you can store that information in a Database instead of in Elasticsearch (the default setting).

In a recent blog post I explained why APIs used to be completely frozen once they were published, and how we have loosened that restriction for Public APIs. Similarly, we did not allow Client Apps to be changed and then re-registered. This was never a good decision, since the Client App does not have anything "connected" to it (the way that an API may). So we should never have restricted the registration of a Client App!

Recently we released version 1.2 of apiman and part of that release includes an effort to rename some concepts to make them more clear (or to better align them with industry standard terminology). Read on below the fold to find out what changed!

The last thing you want after carefully setting up your system with apiman is for someone to be able to call around the gateway and hit your APIs directly. The typical solution for this is to lock down your network so that the only publicly accessible part is the apiman gateway, whilst APIs are hidden in the private part of the network, which apiman can access, but not someone in the outside world. However, in some situations fine-grained network controls may not be available, such as the cloud; or, you may wish to have an additional layer of security to be reassured that no funny business is going on (such as imposters).

The class of solutions to this problem generally falls under the banner of mutual authentication. One such mutual auth offering apiman supports is Mutually Authenticated TLS^[1].

One great advantage of API Management is centralising auth concerns, thereby avoiding burdensome reimplementation issues and streamlining your security processes. The good news is that you can easily configure apiman to handle many common auth use-cases, such as OAuth2 with our popular Keycloak OAuth2 policy which I’ll outline in this blogpost.

If you’re looking to define CORS policies in your API Management layer, then we have an official plugin that should be perfect for the job.

For those unfamiliar with CORS, it’s a way of precisely defining who and how a remote origin may invoke an API’s resources. Generally, due to the same-origin policy, a web browser will only allow the invocation of resources that reside on the same origin as the requesting page. This mitigates a range of malicious script attacks by preventing interaction with remote resources.

However, if we want our resource to be callable by some (or all) other origins, then we need to define a CORS policy which lets user agent know what’s allowed.