Microservices & API Gateways | Mashape

Software developers have been looking at different ways of building software and different architectural patterns have emerged. In this sort of revolution, Microservices and APIs have increasingly assumed greater importance as both a technology and a design pattern. While improving certain aspects of monolithic software, microservices also add additional complexity that has inspired the creation of new design patterns, one of them being the usage of API Gateways. This talk from nginx.conf 2016 highlights the challenges of microservice-oriented architectures, exploring the role of API Gateways and illustrating real-world use cases of their application.
Key Takeaways
- Kong + NGINX provide API Management at unprecedented scale
- Treating your APIs as their own products
- NGINX as the platform for innovation for open-source projects
- The API Gateway pattern
- Microservice orchestration

[@gauravga951]

Very Informative.. Thanks

[@valdezm_com]

Minute 13:57 you mention "up-struc" for mobile clients, creating a composite payload: "collapse all responses in to one response". For some reason the first time I watched this, I thought this meant the gateway had the responsibility of taking different upstreams and composing the response according to some configuration that the Kong(the API gateway) was are of.. But now I understand that what you are saying that you have to create additional single API that is responsible for connecting to the upstreams(services) individually then create some abstraction that will be the composite response payload. That is, I was hoping the API gateway, Kong, would handle these responsibilities and essentially just filter multiple services returning a materialized view.. Was hoping to propagate this to the API gateway as a feature of it, but it isn't.

[@valdezm_com]

Seems like there should be some middleware plugin that could do exactly this. I read somewhere that Kong has response and request plugin, I suppose extending out their functionality is what I am proposing?

[@jaideep_yes]

Thanks

[@stoneshou]

the video mentioned kong stores the information it needs in Postgres or Cassandra. I'm having confusion understanding what Kong needs to store. Isn't it just a bunch of plugin lua scripts on top of nginx?

[@marcopalladino7204]

We are using a datastore to store routing information, plugin configuration, credentials (if you decide to use Kong as a credential store) and so on. Kong tries to optimize database access so ideally, unless you have a very complex use-case, Kong shouldn't make many requests to it.

[@fsa9068]

Nginx-lua doesn't have persistence cache, because of it they use Postgtresql or Cassandra datastore. Kong also caches responses from it

[@richardwang3438]

too many ah...