Develop secure applications on Azure

In this article, we present security activities and controls to consider when you develop applications for the cloud. Security questions and concepts to consider during the implementation and verification phases of the Microsoft Security Development Lifecycle (SDL) are covered. The goal is to help you define activities and Azure services that you can use to develop a more secure application.

The following SDL phases are covered in this article:

  • Implementation
  • Verification

Implementation

The focus of the implementation phase is to establish best practices for early prevention and to detect and remove security issues from the code. Assume that your application will be used in ways that you didn't intend it to be used. This helps you guard against accidental or intentional misuse of your application.

Perform code reviews

Before you check in code, conduct code reviews to increase overall code quality and reduce the risk of creating bugs. You can use Visual Studio to manage the code review process.

Perform static code analysis

Static code analysis (also known as source code analysis) is usually performed as part of a code review. Static code analysis commonly refers to running static code analysis tools to find potential vulnerabilities in non-running code by using techniques like taint checking and data flow analysis.

Azure Marketplace offers developer tools that perform static code analysis and assist with code reviews.

Validate and sanitize every input for your application

Treat all input as untrusted to protect your application from the most common web application vulnerabilities. Untrusted data is a vehicle for injection attacks. Input for your application includes parameters in the URL, input from the user, data from the database or from an API, and anything that is passed in that a user could potentially manipulate. An application should validate that data is syntactically and semantically valid before the application uses the data in any way (including displaying it back to the user).

Validate input early in the data flow to ensure that only properly formed data enters the workflow. You don't want malformed data persisting in your database or triggering a malfunction in a downstream component.

Blocklisting and allowlisting are two general approaches to performing input syntax validation:

  • Blocklisting attempts to check that a given user input doesn't contain "known to be malicious" content.

  • Allowlisting attempts to check that a given user input matches a set of "known good" inputs. Character-based allowlisting is a form of allowlisting where an application checks that user input contains only "known good" characters or that input matches a known format. For example, this might involve checking that a username contains only alphanumeric characters or that it contains exactly two numbers.

Allowlisting is the preferred approach for building secure software. Blocklisting is prone to error because it's impossible to think of a complete list of potentially bad input.

Do this work on the server, not on the client side (or on the server and on the client side).

Verify your application's outputs

Any output that you present either visually or within a document should always be encoded and escaped. Escaping, also known as output encoding, is used to help ensure that untrusted data isn't a vehicle for an injection attack. Escaping, combined with data validation, provides layered defenses to increase security of the system as a whole.

Escaping makes sure that everything is displayed as output. Escaping also lets the interpreter know that the data isn't intended to be executed, and this prevents attacks from working. This is another common attack technique called cross-site scripting (XSS).

If you are using a web framework from a third party, you can verify your options for output encoding on websites by using the OWASP XSS prevention cheat sheet.

Use parameterized queries when you contact the database

Never create an inline database query "on the fly" in your code and send it directly to the database. Malicious code inserted into your application could potentially cause your database to be stolen, wiped, or modified. Your application could also be used to run malicious operating system commands on the operating system that hosts your database.

Instead, use parameterized queries or stored procedures. When you use parameterized queries, you can invoke the procedure from your code safely and pass it a string without worrying that it will be treated as part of the query statement.

Remove standard server headers

Headers like Server, X-Powered-By, and X-AspNet-Version reveal information about the server and underlying technologies. We recommend that you suppress these headers to avoid fingerprinting the application. See removing standard server headers on Azure websites.

Segregate your production data

Your production data, or "real" data, should not be used for development, testing, or any other purpose than what the business intended. A masked (anonymized) dataset should be used for all development and testing.

This means fewer people have access to your real data, which reduces your attack surface. It also means fewer employees see personal data, which eliminates a potential breach in confidentiality.

Implement a strong password policy

To defend against brute-force and dictionary-based guessing, you must implement a strong password policy to ensure that users create a complex password (for example, 12 characters minimum length and requiring alphanumeric and special characters).

Azure Active Directory B2C helps you with password management, by providing self-service password reset, force password reset, and more.

To defend against attacks on default accounts, verify that all keys and passwords are replaceable and that they're generated or replaced after you install resources.

If the application must autogenerate passwords, ensure that the generated passwords are random and that they have high entropy.

Validate file uploads

If your application allows file uploads, consider precautions that you can take for this risky activity. The first step in many attacks is to get some malicious code into a system that is under attack. Using a file upload helps the attacker accomplish this. OWASP offers solutions for validating a file to ensure that the file you're uploading is safe.

Antimalware protection helps identify and remove viruses, spyware, and other malicious software. You can install Microsoft Antimalware or a Microsoft partner's endpoint protection solution (Trend Micro, Broadcom, McAfee, Windows Defender, and Endpoint Protection).

Microsoft Antimalware includes features like real-time protection, scheduled scanning, malware remediation, signature updates, engine updates, samples reporting, and exclusion event collection. You can integrate Microsoft Antimalware and partner solutions with Azure Security Center for ease of deployment and built-in detections (alerts and incidents).

Don't cache sensitive content

Don't cache sensitive content on the browser. Browsers can store information for caching and history. Cached files are stored in a folder like the Temporary Internet Files folder, in the case of Internet Explorer. When these pages are referred to again, the browser displays the pages from its cache. If sensitive information (address, credit card details, Social Security number, username) is displayed to the user, the information might be stored in the browser's cache and be retrievable by examining the browser's cache or by simply pressing the browser's Back button.

Verification

The verification phase involves a comprehensive effort to ensure that the code meets the security and privacy tenets that were established in the preceding phases.

Find and fix vulnerabilities in your application dependencies

You scan your application and its dependent libraries to identify any known vulnerable components. Products that are available to perform this scan include OWASP Dependency Check,Snyk, and Black Duck.

Vulnerability scanning powered by Tinfoil Security is available for Azure App Service Web Apps. Tinfoil Security scanning through App Service offers developers and administrators a fast, integrated, and economical means of discovering and addressing vulnerabilities before a malicious actor can take advantage of them.

Note

You can also integrate Tinfoil Security with Azure AD. Integrating Tinfoil Security with Azure AD provides you with the following benefits:

  • In Azure AD, you can control who has access to Tinfoil Security.
  • Your users can be automatically signed in to Tinfoil Security (single sign-on) by using their Azure AD accounts.
  • You can manage your accounts in a single, central location, the Azure portal.

Test your application in an operating state

Dynamic application security testing (DAST) is a process of testing an application in an operating state to find security vulnerabilities. DAST tools analyze programs while they are executing to find security vulnerabilities such as memory corruption, insecure server configuration, cross-site scripting, user privilege issues, SQL injection, and other critical security concerns.

DAST is different from static application security testing (SAST). SAST tools analyze source code or compiled versions of code when the code is not executing in order to find security flaws.

Perform DAST, preferably with the assistance of a security professional (a penetration tester or vulnerability assessor). If a security professional isn't available, you can perform DAST yourself with a web proxy scanner and some training. Plug in a DAST scanner early on to ensure that you don't introduce obvious security issues into your code. See the OWASP site for a list of web application vulnerability scanners.

Perform fuzz testing

In fuzz testing, you induce program failure by deliberately introducing malformed or random data to an application. Inducing program failure helps reveal potential security issues before the application is released.

Security Risk Detection is the Microsoft unique fuzz testing service for finding security-critical bugs in software.

Conduct attack surface review

Reviewing the attack surface after code completion helps ensure that any design or implementation changes to an application or system has been considered. It helps ensure that any new attack vectors that were created as a result of the changes, including threat models, has been reviewed and mitigated.

You can build a picture of the attack surface by scanning the application. Microsoft offers an attack surface analysis tool called Attack Surface Analyzer. You can choose from many commercial dynamic testing and vulnerability scanning tools or services, including OWASP Zed Attack Proxy Project, Arachni, Skipfish, and w3af. These scanning tools crawl your app and map the parts of the application that are accessible over the web. You can also search the Azure Marketplace for similar developer tools.

Perform security penetration testing

Ensuring that your application is secure is as important as testing any other functionality. Make penetration testing a standard part of the build and deployment process. Schedule regular security tests and vulnerability scanning on deployed applications, and monitor for open ports, endpoints, and attacks.

Run security verification tests

Secure DevOps Kit for Azure (AzSK) contains SVTs for multiple services of the Azure platform. You run these SVTs periodically to ensure that your Azure subscription and the different resources that comprise your application are in a secure state. You can also automate these tests by using the continuous integration/continuous deployment (CI/CD) extensions feature of AzSK, which makes SVTs available as a Visual Studio extension.

Next steps

In the following articles, we recommend security controls and activities that can help you design and deploy secure applications.