Applying Platform Engineering Principles to Network & Infrastructure

Platform engineering holds the promise of delivering on this goal — but it won’t work if it doesn’t include networking, which is key to the entire infrastructure. Without more automated networking operations, the platform engineering won’t deliver the agility of the cloud operations model.

Networking must be used to connect any domain, whether it’s traditional enterprise, cloud, or hybrid. Multi-cloud, automated networking will be needed to deliver the connectivity needed to deliver modern, cloud-native applications along with traditional enterprise apps — but unfortunately networking is often left out of the discussion in the infrastructure planning stages.

In this research brief, we’ll outline the key steps that are needed to deliver networking components to platform engineering, as well as practical ways to get started.

We live in a world of silos. IT assets include portfolios of cross-domain resources, infrastructure, and software. Public cloud infrastructure can become islands unto themselves, with different tools and systems in each cloud. When extending enterprise assets to the cloud world, one is confronted with diverse management tools, standards, and interfaces.

The DevOps revolution ushered in the era of continuous CI/CD, in which code was systematically managed in a pipeline. Translating this to deployment across hybrid infrastructure requires standardizing the components. In a single public cloud infrastructure, this makes sense, but what if you are deploying across multiple infrastructures and multiple clouds? Things get more complicated.

Networking remains one of the challenges bridging these gaps. Development teams can be isolated from infrastructure teams, and they don’t always understand the nuts and bolts of networks (IP addresses? VLANs? BGP?). However, networking remains crucial to connecting the applications and data whether those reside in LANs, WANs, datacenters, Software as a Service (SaaS) applications, or cloud services. These elements need to be connected and managed in a sensible, distributed way.

The process of integrating networking with platform engineering may be daunting, but let’s take a look at ways it can happen.

The DevOps movement was great for accelerating the velocity of application and service development. But in a hybrid enterprise environment, deployment of infrastructure is another story.

A common complaint in IT organizations is that security, infrastructure, and applications teams operate in silos, without understanding each others’ needs. The same challenge can be found in line of businesses (LoBs), which spawned the challenge of “shadow IT.” A LoB might be developing an app without any knowledge of what is used by other teams or partners. For the goals of the organization, this is not always the best approach.

As shown below, this is how typical DevOps silos might evolve, with monolithic applications being built with the resources available to that specific team. For example, Team A might be using public cloud service A, while Team B is using public cloud service B.

One way to tear down the silos is to build a standardized set of infrastructure that can be shared by everyone. Developers need an easy way to provision, configure, and manage infrastructure.

This has the benefit of abstracting away some of the infrastructure complexity from the app, LoB, or DevOps teams. Today, this is largely executed using manual ticket systems and “swivel chair” operations. By integrating platform engineering with support platforms such as ServiceNow, it is possible to build an automation platform to deliver services on demand.

In this model, the standardized infrastructure and platform are abstracted out and made available to LoB and DevOps team as a platform. Practitioners can choose from a set of standard services in a self-service model.

Providing Consistent APIs for Integration & Automation

APIs are the building blocks of automation. They enable systems to be connected and orchestrated using code. To build a set of standardized infrastructure — the platform — APIs play a key role by exposing service configurations. Again, following the cloud model, the key to infrastructure agility is to build commoditized infrastructure hardware that can be programmed via APIs.

Infrastructure and network APIs can be used in platform engineering to define the infrastructure, configuration, and tools available to software developers to build applications for the infrastructure that is available. Network APIs can also be used to program intent of state, by enabling applications to extract analytical and statistical information from network components to support customer experience-based technology use cases.

Integrating Infrastructure as Code (IaC) Into Dev Pipelines

The IaC movement is tightly connected to platform engineering. By using IaC tools, processes for integrating infrastructure deployment can be built directly into applications, via low-code or even no-code automations.

Platform engineering may represent the second wave of IaC. The initial wave of IaC included the use of scripting, coding, and automation tools to drive automation into development pipelines. For example, these might include Jenkins for testing and building products, or Ansible for automating repeatable configuration tasks. More recently, IaC automation has grown more sophisticated with tools such as Python, Pulumi, YAML, Terraform, or Itential — all of which enable infrastructure configuration and orchestration needs to be built directly into an application using low-code or no-code integrations.

A common method of service implementation might include the development team sending service tickets to the infrastructure team to get what they need. What dev teams really need is the capability to automate these service requests, using IaC in conjunction with platform engineering.

Standardizing Networking Elements, Configuration, & Product Requirements

Building infrastructure programmability directly into applications will be one of the keys to achieving success through platform engineering. The platform leader we spoke to said that managing complex, multi-vendor environments is one of the largest challenges that can be aided by automating networks with platform engineering.

“The network industry hasn’t been doing us any favors,” said the solutions engineer. “It’s a mess. There is a lack of standards, different vendors. You have to define your own requirements. We need to standardize.”

The arrival of the scalable commercial off the shelf hardware (COTs) movement enabled clouds to scale by standardizing the infrastructure. Platform engineering can do the same for more complex enterprise environments by providing standardized services and features — and then automating their connectivity via APIs and low-code automations.

By standardizing tools and processes, an organization can present a uniform and consistent set of infrastructure services. This can also simplify management and support by reducing complexity. In many DevOps environments, standardization may be limited to specific teams. By implementing a platform engineering approach to networking, the same infrastructure and networking services can be available to anyone.

Providing On-Demand, Self-Service Infrastructure

The goal of platform engineering with networking is not only to simplify the infrastructure and services available across all times, but also to speed up the deployment of applications and services. Imagine the same “self-service” approach that is available to consumers on the Web, only it’s how enterprise applications and networking services are consumed by end users.

Platform engineering for networking can help IT and networking staff tame some of the complexity of multivendor and non-standard infrastructure using low-code and no-code automation to shield this complexity from end users. The DevOps teams, meanwhile, can build networking into a platform as part of a menu of self-service services.

Like every IT project, platform engineering is likely to be political as well as technical. Implementing a new process involves gaining the support of stakeholders and management. As our bank platform engineer leader told us, it’s as much about people and processes as it is about technology.

The stakes are high: Platform engineering has the potential to not only streamline and improve IT systems, but it can also improve business processes by adding organizational agility and efficiency.

Changing the Operational Mindset

Changing the operational mindset of an organization is the key to success in delivering better service and automation via the platform approach. New operational mindsets can be established by altering the established workflow and setting up a standardized operational model that can enable more automation.

For example, in cloudscale development, CI/CD created an operational mindset built around the speed of development and deployment. The appeal of cloud wasn’t just economy of scale — it was about getting software and services to market faster.

Networking is often perceived as being behind the curve in cloud-native development, but that is due to more complexity in networking environments. The heterogeneous nature of many enterprise networking environments makes standardization more difficult. However, the advent of IaC, APIs, and network automation tools is changing all of that.

Driving more automation into the networking infrastructure will require several changes to operational mindsets, including changing existing network management processes. These changes will require a shift toward network automation, whereby the networking infrastructure is driven by programmable, API-driven software and hardware.

Keys to success include the following:

• Deploying standardized infrastructure, abstracted management, and driving continuous operations with software-based automation.
• Moving to an “API-first” mindset, rather than depending on CLIs and proprietary management models
• Delivering services and infrastructure using a self-service model

More of the details on the changes needed for operational mindset are outlined below.

Roles & Skills Alignment Across Infrastructure & Operations

Developers need infrastructure to run their applications and services. Traditionally, companies have used central infrastructure teams that provision and manage infrastructure on behalf of developers, but this model is prone to bottlenecks as developer requests for infrastructure overwhelm central teams. As modern development teams have taken responsibility over owning and operating their own infrastructure, they also need simple and fast ways of provisioning it while adhering to best practices.

Regardless of implementation details or specific methods, there are some simple requirements that many platform engineering teams follow. All of these requirements are driving toward maximizing the use of the cloud at scale across an organization while being secure and compliant and enabling developers and teams to ship faster.

Simple and powerful user experience: Build curated experiences that empower developers by meeting them at their level of expertise. Use a variety of approaches to provide an ideal user/developer experience, including infrastructure code libraries (reusable pieces of code), infrastructure CLIs, internal developer portals (IDPs), or shared IaC templates.

The process of building a platform engineering approach for networking sounds daunting and “big,” but it doesn’t necessarily have to be this way. You can start small and build from there. The benefit of using APIs, automation tools, and modular services is that they can be phased in gradually, over time.

Reverse the Flow of Rivers

The Itential Automation Platform is based on a fundamentally unique approach to infrastructure and network orchestration. It is designed to enable IT, applications, and network teams to collaborate on building and designing workflows for end-to-end orchestration of infrastructure deployment, provisioning, and management and deliver these services as self-service capabilities to end users.

Itential’s powerful API first capabilities can expose a range of actions to Platform Engineering teams to enable them to easily discover and incorporate network orchestration and automation into their pipelines and applications.
With IAP, users can expose workflows to end users, enable secure, scalable access to onboarded automation assets such as Python, Anisble, etc and provide secure access to API routes for any systems integrated with Itential.

• Low-code workflow builder for orchestrated provisioning of infrastructure services & IT applications
• Securely expose automations for self-service by other teams.
• Expose REST APIs to programmatically invoke automation and orchestration capabilities.
• Publish APIs for DevOps CI/CD pipeline integration.
• Stateful orchestration of operational and execution data, including state of provisioned resources.

This approach to leveraging a network automation and orchestration platform can help accelerate automation and put companies in the direction of making networking part of the platform engineering approach.

Explore the Itential Automation Platform

A platform engineering approach has the potential to guide digital transformations and evolve an organization’s IT approach toward a cloud-native model, with improvements across technology and business operations. Tying networking into platform engineering will be key to its success.

Platform engineering has entered the picture because it enables organizations to speed up, standardize, and improve application and service deployment. Trends such as AI will only increase this need, as the scale and velocity of data and application generation increases. Using platform engineering to accelerate an organization’s capability to compete will become a key differentiator.

Platform engineering is about more than how it climbs the hype cycle, however. It’s a way for engineering teams, including networking, to think about how to automate and improve their processes, including breaking down operational silos. It’s also about empowering the end users across any department — by making self-service access to networking and infrastructure services available to anyone.