Top Considerations for Selecting an API Development Model

Overview

Today, many API management platforms are available in the market. They provide similar features and many options for deployment to meet an organizations’ needs for performance, cost, security, regulatory compliance and infrastructure management.

In this blog, we will discuss the three most common API management deployment models—SaaS, On Premises, and Hybrid—and the use cases to consider for selecting one over the other along with their pros and cons. This blog will be of particular interest to CIOs, CTOs, technical architects and product managers who are looking to set up an API management platform for their digital transformation initiatives.

The SaaS Deployment Model

In this model of deployment, the entire underlying infrastructure along with all the components of the API management platform are hosted on the public cloud and maintained by the service provider. The organization needs a license or subscription for the services and is ready to start working on the platform without managing any physical infrastructure.

Potential Use Cases

The SaaS solutions can be leveraged by all types of businesses. They are, however, more attractive to smaller businesses and startups (for example, a small manufacturer that wants to set up a tracking system for the maintenance of different plants in one or more countries). With a SaaS solution, an organization only needs to focus on services that are to be exposed through APIs vs having to invest in infrastructure, security, and other basic requirements of an API management platform.

The On Premises Deployment Model

This deployment model requires the API management platform infrastructure to reside in data centers or on a private cloud. Organizations have the freedom to customize the platform and they exercise complete control over the infrastructure, including decisions about the data and components to be installed on the platform.    

Potential Use Cases

Organizations dealing with personal, financial and sensitive data, regulated by data protection laws and    government regulatory compliance prefer on premises deployment model. Banks and other financial institutions (for example, insurance companies) as well as healthcare service providers tend to opt for this type of API deployment model.

The Hybrid Deployment Model

The Hybrid deployment model provides the best of both Cloud and On Premises deployment features. In this model, the runtime components of the API management platform are installed on premises or on a private cloud infrastructure, while the management/control plane components of the platform are hosted and managed by the platform provider on their cloud. This allows APIs to be deployed on a cloud platform that is preferred by the API provider and collocated with the backend services.  

Potential Use Cases

Large organizations that need to expose both sensitive and non-sensitive data may find the Hybrid deployment model more suitable. For example, banking, financial, or healthcare institutions who provide both sensitive and non-sensitive data to their partners and end consumers may prefer to expose their core  services on an internal network only, while exposing other services externally to third-party providers on the cloud. 

In Conclusion

It is clear, when selecting the right deployment model for your enterprise, you need to carefully weigh the pros and cons. Each of these models could be suitable for any organization based on their specific requirements, regulatory policies, security, scalability cost and other factors. While a Cloud-based deployment model provides scalability and faster go-to-market product delivery, and the On Premises deployment model offers more customization and control, a Hybrid deployment model can offer the best of both. When choosing a deployment model for your organization, also consider the API management platform your organization utilizes or plans to migrate to. All leading API management platforms like Apigee, MuleSoft, IBM API Connect, Kong, and Axway can support the three deployment models covered in the blog.