DevOps Blog

Stay in the know with in-depth articles about DevOps, micro-services and cloud native topics, delivered to you weekly.

Cloud-Native API Gateway Services

Written by Marius Rimkus
on March 04, 2020

As the cloud continues to become a staple feature in modern enterprises, there is renewed interest in determining whether your workloads – and the overall organization – are ready to operate in a cloud-native environment.

As there are various considerations involved in this, organizations need to carefully analyze everything from the vendor to the orchestration tool.

Here, we are going to delve into the details of one such consideration: cloud-native API gateways. As we’ve already introduced API gateways in “The Complete Overview of DevOps Cloud-Native Tools Landscape,” this piece will take a deeper look into how they work.

 

What Is An API Gateway?

The exact definition of an API gateway is hard to pin down; people use various terms to define its impact, such as a single point of entry, an interface and reverse proxy.

Nonetheless, the idea remains the same – an API gateway functions as a layer that separates your clients and services. While the conventional method meant that clients directed their requests to the microservices directly, the requests are now handled by the gateway which then passes the requests on to the appropriate service.

As the de facto client request coordinator and orchestrator, API gateway simplifies the implementation of a distributed system. It can be particularly useful for organizations trying to shift from monolithic system architecture to microservices.

 

What Is The Use Of An API Gateway?

One of the main reasons for the increasing popularity of API gateways is the simultaneous rise in the use of microservices software.

Contrary to the standalone monolithic design, microservices divide the entire application into various loosely connected services. Each of these microservices is self-sufficient and possesses distinct functionality. However, this vital feature is what evokes the need for an API gateway.

Don’t get us wrong – microservices provide a vast array of benefits as they make development, deployment, and maintenance incredibly easy. But when it comes to request processing, having several microservices makes it hard for the client to access a particular service quickly.

This is where the API gateway comes in; by acting as the reverse proxy, it solves the client-side problem and allows easy access to various microservices. Using the API server, the client only has to send a solitary request.

It also consists of an HTTP server that has routes associated with each microservice. Upon receiving the request, the gateway immediately maps the entire request and identifies the relevant service required to deliver the request.

In fact, it can even map the request parameters to the input arguments required by a specific microservice. Some of them require input argument validation as a necessary prerequisite to process the request, and the API gateway can make sure no human intervention is required for this step.

This concludes routing – the primary function of an API gateway.

 

What Other Functions Can An API Gateway Perform?

Although the importance of APIs to modern organizational applications cannot be emphasized enough, too many can prove to be a management headache.

API gateways offer users a large number of management features that help with core tasks such as security. Here’s an extensive list of features:

Authentication

API authentication is an important security check that is conventionally required when accessing each individual microservice. With an API gateway in place, the calls can be authenticated at the gateway itself.

This way, even if the client requests access from multiple microservices, the gateway will ensure that the API call is authenticated only once. This drastically reduces latency issues and provides consistent authentication checks throughout the application.

Metric Collection

Developers have long wished for analytics to be collected where requests are funneled since this provides the perfect location to process all requests. With an API gateway in place, you are provided a comprehensive ‘gateway’ where all processes will inevitably go through.

Utilizing this analytical capability provides users with a deeper insight and, ultimately, more control. For instance,  you can analyze the number of requests being forwarded by a particular user or the number of requests being relayed to individual microservices through the gateway.

In return, you can use this information to implement rate-limiting. This allows you to reject users that are making excessive requests so as not to overburden the microservices.

Response Transformation

There are various devices that access modern-day applications, ranging from highly resourceful computers to smartphones and tablets – and so, each user requires access to different information and at different requirements.

For instance, mobile users require less data than client requests that originate from desktop systems. Since API gateway is the only source of entry, it knows which client requires what.

This information can be subsequently used to make an internal call and give the data to different clients as per their requirements.

Input Validation

API gateways also possess the ability to process simple logic. Take input validation, for instance, where the gateway verifies that the request contains all the necessary data before it is sent to the microservices.

Additionally, it also checks the request for the correct format and code so that it reaches the service in a ready-to-execute state.

Load Balancing

You need to ensure that no node is overburdened with requests so that your application continues to function smoothly. An API gateway can be programmed to be smart enough to balance loads across different nodes present in a particular microservice.

By monitoring the entirety of the load, it works as a highly efficient load balancer.

Service Discovery And Requests Dispatching

This is yet another feature that stems from the fact that an API gateway acts as a reverse proxy for clients to access microservices.

In a typical process, it will receive a request and deliver it to all required microservices. Once they respond with the answer, the gateway combines the results before they reach the client.

During this process, the gateway records the basic response time taken by each node of a particular microservice – this can come handy when segmenting API calls based on priority. For high priority calls, the request can be routed to the node that responds the fastest.

Circuit Breaker

Just like a conventional circuit breaker, this metaphorical one also closes down an affected circuit until the issue is resolved.

Here’s what you need to know; component failure under data load or due to any other issue is not a rarity when it comes to applications. To deal with a situation like this, API gateways utilize a ‘circuit breaker’ methodology.

Once a specific request threshold is met, the failed component ceases to receive any more data from the gateway. Not only does this buy time to analyze logs and push a fix via an update, it effectively closes the entire circuit down until the core problem is fixed.

What Are The Benefits Of An API Gateway?

If the added functionality offered by an API gateway wasn’t enough to convince you, then let us tell you that they also come with various other benefits.

When put into perspective, these features allow you to package your loosely-coupled microservices into a concise application. Here are the benefits afforded by an API gateway:

A Cleaner Client-Side Code

With the client and backend services decoupled, the client isn’t required to know how each microservice has been composed. With API gateway handling all requests, code maintenance becomes much simpler.

This also negates the need for individual microservices to keep track of endpoints or deal with request failures. This saves time as developers aren’t required to code ‘logic’ in their applications, resulting in enhanced consumer experience.

Decreased Latency

Even a single request from the client application can require API calls to multiple microservices at the same time.

Take Amazon for example; one simple search is likely to return a list of various products, their stock availability, and reviews. If not for an API gateway, the client request will take multiple trips to each microservice. The result is a significant increase in latency and, consequently, poor user experience.

Enhanced Application Security

If a developer chooses not to include an API gateway, then they have to enable each microservice to conduct its own security-related analysis about the client requests they encounter.

For the developer, this translates to hours’ worth of additional coding to add this complex decision-making skill to the code. This is necessary because, without a single point of entry, each microservice has to answer the following questions:

  • Is the request authorized?
  • Is the communication channel secure?
  • Is the request complete in all of the required parameters?
  • Is the client issuing too many requests?

Answering all these questions inevitably makes the code more intricate. Consequently, the more complicated it is, the more chances there are of ‘bugs.’

From an operator’s perspective, a complex code requires increased maintenance and downtime. Not only does this translate to a severely affected user experience, but it also increases the overall attack surfaces available to bad actors.

All of these problems can be avoided with a centralized point of entry. An API gateway ensures all authentication and security checks are handled in one place, leaving the backend microservices unaffected for smooth user experience as well as increased security.

Popular API Gateways

There are various open source API gateways available in the market. Here is a list of the most popular ones that are being used by thousands of developers worldwide:

Kong

As one of the first cloud-native API gateways, Kong is a popular platform in this domain. Written primarily in Lua, the gateway is augmented by both open source and proprietary plugins.

Similar other tools in the list, Kong comes with an integrated, same-vendor API management service – the Kong Service Control platform.

Tyk

An open-source API gateway, Tyk is developed in Go that allows it to be compatible with plugins written in different coding languages.

What’s different about Tyk is their management service; the company offers its proprietary API management platform in two different options – as an on-premise installation or a managed service with a fee.

Ambassador

Yet another API gateway, the Ambassador is native to Kubernetes. It is built on top of the Envoy server that serves as an ingress controller.

If your microservices structure is based on a Kubernetes cluster, then using Ambassador is a sensible choice due to its inherent integration and the ability to efficiently run YAML configuration files.

Express Gateway

Based on Node.js, Express Gateway is completely built on Express and Express Middleware platforms.

Its simplistic configuration process and the ability to support JavaScript plugins are one of the unique points that distinguish it from other gateways.  It can also reuse existing Express Middleware functionality for greater effect.

 

EndNote

To conclude, API gateway is a reverse proxy that allows microservices to be offered up as APIs. The biggest advantage of this is that it helps negate the possibility of a security breach by exposing back-end microservices to the client’s request.

Acting as the central interface, not only does it allow developers to keep their codes relatively simpler but also reduces latency issues. With a more cohesive and secure microservice structure, you are able to provide an enhanced user experience to the client.

With a wide range of options available, the best API gateway platform ultimately depends on the architecture of your system and the nature of your workloads.