Viewing Your World Through Security Coloured DevOps Glasses

[rt_reading_time label=”Read Time:” postfix=”minutes” postfix_singular=”minute”]

Imagine a hypothetical organization, in a utopian world, where the security team works alongside operations, developers and business owners; collaborating, sharing goals, coding and using the same tools. In this organization, security is embedded throughout the value supply chain:

  • Service design including threat modeling
  • User stories include security acceptance criteria
  • Security engineers listening in instead of stopping changes
  • Security policies and standards are translated into code
  • Developers and operations can interact with the security services using API
  • Security test frameworks are created by the security team, and used as part of the delivery pipeline
  • Security engineers create safe code to handle untrusted input
  • Security events are collected, aggregated and correlated, leveraging data science capabilities to make informed decisions

This utopian organization is very difficult to grasp, but why is that? Basically, security is typically considered an afterthought, where risk assessments are cumbersome tasks executed annually, coupled with penetration testing. The goal is to comply with minimal regulations or certifications required by governments and customers, creating a wide window of exposure, and only covering a small portion of the services.

For Agile teams, traditional security represents a waterfall model, the opposite of Agile, where the security organization’s responsibility is to define policies and standards that are almost impossible to translate into the Agile processes.  For this reason, security usually ends up applying at the perimeter, mostly focused at the infrastructure (e.g. firewalls, IDS/IPS, WAF, SIEM) but very far from the applications or services.

This model is far from optimal. One of the biggest problems of the security field is that failure to provide an adequate level of protection can cause huge damage to organizations and its customers, in the worst cases leading to massive loss of trust and headlines in the news such as:

Having the right level of security is hard to achieve and budget constraints don’t make it any easier. With Agile development and DevOps movements increasing the rate of change, the gap between the security teams desired state and actual state will increase. However security teams can be part of this DevOps cultural movement, helping developers and operations to deliver incremental, efficient and secure features to customers.

DevOps is a cultural mindset, where operations works alongside developers, sharing goals and tools, collaborating and embracing change. DevOps promotes incremental changes, creating a value supply chain process that allows changes to be introduced at a rapid pace, while providing confidence. Adding security to this cultural mindset is a must to modern and agile organization, realizing that security must be integrated into their current teams and processes.

Adding security to the DevOps movement—DevSecOps, or Rugged Software—allows development, operation and security teams to work together, with wide communication channels. Understanding that security is everyone’s goal, they can then embrace change with a security mindset, and look for new and Agile methods to solve security problems.

How teams can be organized?

Security must be embedded into the development and operation processes. For this reason, every Agile team should have a security engineer (basically, an engineer who understands development and security) supporting the team with security coding practices, translating the security policies and standards into code, and creating a feedback loop from the Agile team to the security organization in order to improve those policies and standards.

This security engineer can be part of multiple Agile teams, participating in application design discussions, sprint planning, review and retrospective meetings, and specific stand-ups depending on the user stories that are part of the sprint. The risk associated with these user stories could help to define the level of participation of the security engineer, where user stories that implement functionalities with medium or high risk (e.g. data moving across trust boundaries (user submitted input), elevation of privileges. storing/retrieving PII (Personally Identifiable Information), PHI (Protected Health Information) or credit card data), might require more participation, either providing guidance on the security controls to be implemented or creating security tests to evaluate the effectiveness of the implemented security controls.

This approach increases the security awareness and security coding standards of the team, while gradually instilling a security mindset on the members of the Agile team. Also, just as operations must be a part of the Agile team (describing and implementing the infrastructure-as-code) the security operations team can be a part as well, implementing the security controls at the infrastructure level as code using the same tools as the operations team: configuration management, defining container images, and writing network security policies as code (Ansible or API driven). Basically, security checklists are translated into code by the security teams, working together with operations. Security infrastructure tests can also be added to provide confidence over the security controls.  

How to integrate security with the Agile processes?

An Agile software development lifecycle is an iterative process, using the Microsoft Security SDLC and scrum, security could be integrated as follows:

Requirements gathering: Include security requirements as part of non-functional requirements.

System design: Use threat modeling to identify, quantify, and address the security risk associated with an application.


  • Backlog Grooming: Review/update threat model, define security acceptance criteria for user stories, define security tests as part of user stories
  • Sprint Planning: Explain security acceptance criteria to the team before committing to the user stories to be implemented during the sprint
  • Sprint: Perform security static analysis of the code, implement security tests including fuzzing or dynamic analysis based on the application domain, vulnerability dependency check (check vulnerabilities in libraries and other dependencies, as well as the use of approved/trusted dependencies). These tasks can be implemented as part of the CI/CD
  • Sprint Review: Share results of security tests and controls implemented
  • Sprint Retrospective: Use collected data and team self-evaluation to improve the process


  • Perform continuous dynamic analysis tests on the production platform in order to identify security gaps
  • Collect and correlate events, aggregating data from multiples sources, providing greater visibility and allowing informed decision
  • Develop/improve Incident Response Plan

Having an updated threat model of the application is very important. This will allow the team to apply the right security controls to mitigate the identified security threats. Implementing TLS and Deep Packet Inspection firewalls won’t protect the application from SQL injection attacks; only input validation, applying whitelisting or input sanitization coupled with SQL prepared statements, will provide an adequate level of protection.

As the code moves through the different stages of the CI/CD pipeline, security controls can be introduced to test the security of the changes. Checking for known vulnerabilities in the application’s dependencies can be done with the help of tools like the OWASP Dependency-Check Project, in addition, having a list of tested and approved dependencies that are constantly updated and enforced in the CI/CD helps to mitigate typo-squatting attacks (injecting malicious packages into known registries with names very similar to existing packages).

Vulnerability scanning tools can be introduced as part of the security testing in the CI/CD. It’s important to have an updated vulnerability database with alerting capabilities to notify the team if the database failed to update, while also adding the current database release date to the testing report.

Security policy enforcement and reporting is a must for highly regulated companies. Requirements like strong authentication, use of Multi-factor Authentication (MFA), and data encryption at rest and in transit can be achieved working together with the operations team. The use of cloud platforms also help to simplify this process, where policies are translated directly into code.

MFA enforcement and reporting can be achieved in cloud platforms like AWS using IAM policies coupled with the AWS Config service, using rules to trigger alarms or specific actions in response to changes to the configuration. For example, EC2 instances can be tagged with data classification values (public, private, secret, etc) based on the data being processed or stored, then with specific AWS Config rules, it’s not only possible to check the existence of this tag, but also validate that volumes are encrypted where the data classification value is equal or greater than “secret”. Using the same concept, AWS Config rules can be created to check the use of TLS v1.1 or v1.2 as the only available policies in the load balancers or check S3 bucket permissions to prevent public access.

Adding a Blue/Red team approach to the above process would improve the security of the services and the organization as a whole. Having a team continuously challenging the security controls and attempting different attack vectors to gain access to the resources, fits well with the Agile mindset.


In order to be effective, security can and must be a part of all of the organization’s processes. Leaving security as an afterthought leads to higher security risks, while enforcing security policies and standards without teaming-up with the rest of the organization creates more friction and insecure services. Security must embrace the DevOps mindset in order to keep up with the rapid pace of change and continuously add value to the customers. While there is no one solution or recipe that can fit all organizations, applying Agile values and practices is a good starting point.

Hidden layer

Share on linkedin
Share on twitter
Share on facebook
Share on email

Onica Insights

Stay up to date with the latest perspectives, tips, and news directly to your inbox.

Explore More Cloud Insights from Onica


The latest perspectives on navigating an ever-changing cloud landscape

Case Studies

Explore how our customers are driving cloud innovation in their industries


Watch an on-demand library of cloud tutorials, tips and tricks


Learn how to succeed in the cloud with deep-dives into pressing cloud topics