‘Security by design’ and adding compliance to automation
By Jason McKay, CTO and SVP of Engineering, Logicworks
Security is “job zero” for every company. If you are putting your customers or users at risk, you will not be in business for long. And that begins with taking a more proactive approach to infrastructure security — one that does not rely on the typical protective or reactive third party security tools, but builds security into your infrastructure from the ground up.
As your company moves to the cloud, it has an opportunity to start fresh and rethink who and what is responsible for security in your environment. You also want to be able to integrate security processes into your development pipeline and maintain consistent security configurations even as your applications constantly change. This has led to the rise of Security by Design.
The security by design approach
Security by design (SbD) is an approach to security that allows you to formalize infrastructure design and automate security controls so that you can build security into every part of the IT management process. In practical terms, this means that your engineers spend time developing software that controls the security of your system in a consistent way 24×7, rather than spending time manually building, configuring, and patching individual servers.
This approach to system design is not new, but the rise of public cloud has made SbD far simpler to execute. Amazon Web Services has recently been actively promoting the approach and formalizing it for the cloud audience. Other vendors promote similar or related concepts, often called Secure DevOps or Security Automation or Security-as-Code or SecOps. The practice becomes more important as your environment becomes more complex, and AWS actually has many native services that, if configured and orchestrated in the right way, create a system that is more secure than a manually-configured on-premises environment.
Does this mean that companies no longer need security professionals, just security-trained DevOps engineers? Not at all. When security professionals embrace this approach, they have far greater impact than in the past. This is actually an opportunity for security professionals to get what they have always dreamed of: introducing security earlier in the development process. Rather than retroactively enforcing security policies — and always being behind — they are part of the architecture planning process from Day 1, can code their desired specifications into templates, and always know that their desired configurations are enforced. They no longer need to be consulted on each and every infrastructure change, they only need to be consulted when the infrastructure templates change in a significant way. This means less repetitive busy-work, more focus on real issues.
Security by design in practice
In practice, SbD is about coding standardized, repeatable, automated architectures so that your security and audit standards remain consistent across multiple environments. Your goals should be:
- Controlled, standardized build process: Code architecture design into a template that can build out a cloud environment. In AWS, you do this with CloudFormation. You then code OS configurations into a configuration management tool like Puppet.
- Controlled, standardized update process: Put your CloudFormation templates and Puppet manifests in a source code management tool like Git that allows you to version templates, roll back changes, see who did what, etc.
- Automated infrastructure and code security testing as part of CI/CD pipeline: Integrate both infrastructure and code-level tests into code deployment process as well as the configuration management update process. At Logicworks, we often use AWS CodeDeploy to structure the code deployment process. You can also use Docker and AWS ECS.
- Enforced configurations in production: Create configuration management scripts that continually run against all your environments to enforce configurations. Usually hosted in a central management hub, and necessitates a hub-spoke VPC design approach.
- Mature monitoring tools with data subject to intelligent, well-trained human assessment: In compliant environments, your monitoring tools are usually mandated and logs must be subject to human review; we use native AWS tools like AWS CloudWatch, CloudTrail, and Inspector, as well as Alert Logic IDS and Log Manager and Sumo Logic to meet most requirements. SumoLogic helps us use machine learning to create custom alerts that notify our 24×7 Network Operations Center when unusual activity occurs, so that those engineers can take appropriate action with more accurate real-time data.
- Little to no direct human intervention in the environment…ever: Once all these tools are in place, you should no longer need to directly modify individual instances or configurations. You should instead modify the template or script to update (or more ideally, relaunch) the environment.
We have gone into significant technical depth into Logicworks’ security automation practices in other places; you can see our Sr. Solutions Architect’s talk about security automation here, watch him talk about our general automation practices here, or read this in-depth overview of our automation practices.
Here are some other great resources about Security by Design and Secure DevOps:
- AWS Security by Design White paper
- SANS Institute: Continuous Security: Implementing the Critical Controls in a DevOps Environment
Compliance + security by design
As you can imagine, the SbD approach has significant positive impacts on compliance efforts. The hardest thing to achieve in infrastructure compliance is not getting security and logging tools set up and configured, it is maintaining those standards over time. In the old world, systems changed infrequently with long lead-times, and GRC teams could always spend 2-3 weeks evaluating and documenting change manually (usually in a spreadsheet). In the cloud, when code gets pushed weekly and infrastructure is scalable, this manual compliance approach can severely limit the success of cloud projects, slow down DevOps teams, and frustrate both business and IT.
Running applications in the cloud requires a new approach to compliance. Ideally, we need a system that empowers developers and engineers to work in an agile fashion while still maintaining security and compliance standards; we need a toolchain that a) makes it easier to build out compliant environments, b) provides guardrails to prevent engineers/developers from launching resources outside of compliance parameters, and c) provides ongoing documentation about the configuration of infrastructure resources. The toolchains we have already described — templating, configuration management, monitoring — allow us to launch new compliant environments trivially, ensures very limited access to the environment and full documentation on every change. Together, this means a greatly reduced risk of undocumented configuration change, error, or lack of adequate knowledge about where sensitive data lives, and therefore greatly reduced risk of compliance violations.
When systems are complex, there must be an equally powerful set of management tools and processes to enforce and maintain configurations. Continuous compliance is only possible if you treat your infrastructure as code. If your infrastructure can be controlled programmatically, your security and compliance parameters are just pieces of code, capable of being changed more flexibly, versioned in Git like any piece of software, and automated to self-correct errors. This is the future of any type of security in the cloud.
The future of SbD
SbD allows customers to automate the fundamental architecture and, as AWS says,”render[s] non-compliance for IT controls a thing of the past.”
Recent announcements out of AWS re:Invent 2016 are particularly exciting. AWS launched a major update to their EC2 Systems Manager tool, which is a management service that helps you automatically collect software inventory, apply OS patches, create system images, and configure Windows and Linux operating systems. Basically, AWS is filling the gaps in its existing SbD toolchain, stringing together a lot of the controls described above and allowing you to define and track system configurations, prevent drift, and maintain software compliance. Although EC2 Systems Manager was upstaged by several more headline-worthy releases, the service will make a significant difference to compliance teams in the cloud.
In the future, expect AWS and other cloud platforms to launch more comprehensive tools that make it easier for enterprises to achieve SbD in the cloud. The tools currently exist; but assembling these tools into a robust framework can be a challenge for most IT teams. Expect enterprises to turn towards security-focused partners to fill the skills gap.
- » Five ways big data and integration enables the factory of the future
- » AWS hits $3.5bn in revenue for Q416, takes on ‘surge’ from Microsoft, IBM, and Google
- » Nutanix, HPE and Cisco top 451 rankings for servers and converged infrastructure
- » Financial services and solving the great cloud conundrum
- » Where is the true state of the cloud in 2017? Analysing two influential reports