How to Build a Feedback Loop That Narrows the Gap Between Development and Security
September 23, 2021

Lebin Cheng
Imperva

There's a tug-of-war happening between the development operations team (DevOps) that's responsible for building the company's innovations and the security operations team (SecOps) that's tasked with keeping everything protected.

The growing proliferation of application programming interfaces (APIs) is further exacerbating the tension between these functions. On one side, you have SecOps teams, who have a difficult time gaining visibility into API transactions as API schemas are often lightly documented and subject to frequent changes. While on the other side, there are DevOps teams, who need the freedom to rapidly adjust APIs to meet changing business needs without the burden of manually updating API specifications for security testing and policy definitions.

Although the SecOps team wants to keep a watchful eye on the behaviors of the dozens or hundreds of APIs operating within their network, today's conventional API security methods slow DevOps and make the organization less efficient.

As the connective tissue that binds modern, cloud-native applications together, APIs are essential; but, they're also introducing more cybersecurity risk to the organization. By 2022, it's predicted that APIs will become the most frequently attacked enterprise web application vector.

To fully realize a successful approach to development security operations (DevSecOps) for API security, creating an effective feedback loop between DevOps and SecOps teams is critical to getting a grasp on API security risks.

Establishing a Feedback Loop that Discovers, Monitors and Secures APIs

Historically, applications were deployed under the assumption they would be protected by the network perimeter. As modern software development moves into cloud-native environments, this traditional concept is less effective and leaves the process exposed to additional security vulnerabilities.

By establishing and implementing a feedback loop between DevOps and SecOps, organizations can streamline application release workflows and enable developers to focus on delivering an optimal digital experience while providing the SecOps team with visibility and control over the application runtime.

The ideal feedback loop should encompass three critical domains: discovery, monitoring and security.

Discovery: Maintain an always up-to-date API inventory with contextual data labels. Ideally, this should be done autonomously with an unobtrusive solution that continuously keeps the API inventory up-to-date with data security classifications. For some, discovery simply entails mapping API service endpoints, but that's not enough. Instead, you need to know what data each API is accessing — and shift to a data-centric approach to API security.

Monitoring: Generate developer-sourced specifications and check against security best practices. Functional or regression testing is monitored to validate specifications or to generate specifications if they're not available. Enabling automation ensures protection can keep pace with application changes without manual intervention. That way, new APIs discovered during runtime are checked in the next cycle while API calls in testing help prepare the runtime model.

Security: For API-first apps, API specification is always completed and updated before actual implementation. For other applications, API specification can be used as a reference, but dynamic discovery is needed to ensure the actual implementation and API specifications are in sync. This also means that stringent positive enforcement is not possible. An automated learning system is needed to build a new baseline every time a new API specification is discovered or updated. The new baseline helps to identify anomalies accurately and drive security policy actions without manual intervention.

This feedback loop gives the SecOps team the visibility they need into potential threats without slowing down the development process. It's analogous to adding a security camera to monitor the production floor of a warehouse. The SecOps team gets visibility around-the-clock, monitors for suspicious activity and can react as soon as something nefarious happens. In a software production environment, AI and machine learning are essential for helping automate this activity and to reduce the time it takes to respond.

What's Needed Longer Term

The idea behind a development security operations (DevSecOps) process is sound, but the approach is often flawed because these two functional departments cannot simply be locked in the same room and expected to exist symbiotically.

Both DevOps and SecOps want a frictionless relationship where developers are enabled to move fast and innovate the business, without putting it at risk. However, both sides lack the tools needed to monitor what's happening in development and production — particularly the API calls between internal and external applications and services.

Through the implementation of the API discovery and risk assessment feedback loop, organizations can successfully streamline the resources needed to manually fill the gap while mitigating security risks.

Lebin Cheng is Head of API Security, Office of the CTO, at Imperva
Share this

Industry News

April 03, 2025

StackGen has partnered with Google Cloud Platform (GCP) to bring its platform to the Google Cloud Marketplace.

April 03, 2025

Tricentis announced its spring release of new cloud capabilities for the company’s AI-powered, model-based test automation solution, Tricentis Tosca.

April 03, 2025

Lucid Software has acquired airfocus, an AI-powered product management and roadmapping platform designed to help teams prioritize and build the right products faster.

April 03, 2025

AutonomyAI announced its launch from stealth with $4 million in pre-seed funding.

April 02, 2025

Kong announced the launch of the latest version of Kong AI Gateway, which introduces new features to provide the AI security and governance guardrails needed to make GenAI and Agentic AI production-ready.

April 02, 2025

Traefik Labs announced significant enhancements to its AI Gateway platform along with new developer tools designed to streamline enterprise AI adoption and API development.

April 02, 2025

Zencoder released its next-generation AI coding and unit testing agents, designed to accelerate software development for professional engineers.

April 02, 2025

Windsurf (formerly Codeium) and Netlify announced a new technology partnership that brings seamless, one-click deployment directly into the developer's integrated development environment (IDE.)

April 02, 2025

Opsera raised $20M in Series B funding.

April 02, 2025

The Cloud Native Computing Foundation® (CNCF®), which builds sustainable ecosystems for cloud native software, is making significant updates to its certification offerings.

April 01, 2025

The Cloud Native Computing Foundation® (CNCF®), which builds sustainable ecosystems for cloud native software, announced the Golden Kubestronaut program, a distinguished recognition for professionals who have demonstrated the highest level of expertise in Kubernetes, cloud native technologies, and Linux administration.

April 01, 2025

Red Hat announced new capabilities and enhancements for Red Hat Developer Hub, Red Hat’s enterprise-grade internal developer portal based on the Backstage project.

April 01, 2025

Platform9 announced that Private Cloud Director Community Edition is generally available.

March 31, 2025

Sonatype expanded support for software development in Rust via the Cargo registry to the entire Sonatype product suite.

March 31, 2025

CloudBolt Software announced its acquisition of StormForge, a provider of machine learning-powered Kubernetes resource optimization.