HTTP Strict Transport Security (HSTS) is a web security policy mechanism that enables web sites to declare themselves accessible only via secure connections. This helps protect websites and users from protocol downgrade and cookie hijacking attacks.
Table of contents
- Why was HSTS introduced?
- How does HSTS work?
- What is an example situation involving HSTS?
- What are the limitations of HSTS?
- What are HSTS deployment best practices?
- What is the history of HSTS?
- What browsers support HSTS?
- How UpGuard can check your and your vendors' security controls including HSTS
Why was HSTS introduced?
HTTP is used over various transports, typically the Transmission Control Protocol (TCP).
However, TCP does not provide integrity protection, confidentiality or secure host identification.
This led to the development of Secure Sockets Layer (SSL) and its successor Transport Layer Security (TLS). SSL/TLS provide an encryption layer between application protocols and TCP, commonly known as HTTPS.
In general, user agents (like web browsers) will employ various local security policies to decide how to interact with a host, based on a negotiation between the server, user preferences and their communication method (HTTP or HTTPS).
However, some user agents allow users to choose to continue to interact with a website when they are unable to establish a secure connection. This could happen when a TLS certificate's trust chain is not validated, when it has expired or when the TLS host's domain name appears incorrectly in the TLS certificate.
This behavior is called click-through insecurity.
While giving users the option to continue to use a website despite a lack of HTTPS can keep users happy, it can introduce attack vectors that leave users open to certain types of cyber attacks, particularly man-in-the-middle attacks (MITM attacks), downgrade attacks and session hijacking attacks.
As HSTS allows websites to declare they are only accessible through a secure connection, they can prevent users from connecting to them over any HTTP connection.
This prevents a security vulnerability known as SSL-stripping.
SSL-stripping is a downgrade attack that was introduced by Moxie Marlinspike in his 2009 BlackHat Federal talk New Tricks for Defeating SSL in Practice.
A downgrade attack is a form of cryptographic attack on a computer system or in this case, a communications protocol that makes it abandon its encrypted connection (HTTPS) in favor of an older, unencrypted connection (HTTP) that is typically provided for backwards compatibility with older systems.
SSL-stripping is implemented as part of a man-in-the-middle attack where web traffic is intercepted and redirected from the secure HTTPS version of the website to an unencrypted HTTP version.
The primary reason this attack continues to be successful is that many websites continue to not use TLS/SSL certificates. This makes it impossible to know (without prior knowledge) whether a website's lack of HTTPS is due to an SSL-stripping attack or because they don't have a TLS certificate.
Additionally, there are no warnings to warn the user during the downgrade process, making the attack hard to detect even for the most vigilant user.
With the creation of a tool by Marlinspike to fully automate this type of attack, it represents a real cybersecurity risk.
Session hijacking or cookie hijacking is another vulnerability that is enabled through click-through insecurity.
Session hijacking exploits a valid computer session to gain unauthorized access to information or services.
This is particularly relevant for web developers as cookies are used to maintain a session on many websites.
If a website does not flag their cookies as Secure, telling user agents to only send cookies over HTTPS, they can be easily stolen by an attacker. As non-Secure cookies are returned to the host regardless of transport security, leaving them open to man-in-the-middle attacks.
Once an attacker has access to the cookies, they can then impersonate the user on a legitimate website.
How does HSTS work?
HSTS enables web servers to declare that any interactions by web browsers and other user agents must be conducted over HTTPS connections and not insecure HTTP connections.
A server can implement an HSTS Policy by supplying a response header over an HTTPS connection (HSTS headers sent over HTTP response headers are ignored). The HSTS header is name "Strict-Transport-Security and also specifies a period of time during which the user agent should only access the service via HTTPS requests.
This means the first time a site is accessed using HTTPS it returns the Strict-Transport-Security header, the browser records this information, so future attempts to load the site using HTTP automatically use HTTPS.
When the expiration time specified by the Strict-Transport-Security header elapses, the next attempt to load the site via HTTP will proceed as normal instead of automatically using HTTPS.
However, whenever the Strict-Transport-Security header is delivered to the user agent, it will update the expiration time for that site, so sites can refresh this information and prevent the timeout from expiring.
Should it be necessary to disable HSTS, web servers can set the max-age to 0 (over a HTTPS connection) to immediately expire the HSTS header, allowing access via HTTP requests.
For example, a server could send a header that requests that future requests for the next year only use HTTPS via Strict-Transport-Security: max-age=31536000
When a web application issues a HSTS Policy to user agents, conforming user agents behave as follows:
- Any insecure links are automatically turn into secure links (e.g. http://example.com/ will be modified to https://example.com before accessing the server)
- If a secure connection cannot be ensured (e.g. the server does not have a valid certificate), the user agent will terminate the connection and not allow the user to access the website.
The most important thing to understand is that a HSTS Policy prevents click-through insecurity by not allowing the end user to use the insecure connection.
What is an example situation involving HSTS?
Imagine your staff member uses a free WiFi access point at a cafe and starts surfing the web, visiting your organization's payroll system.
Unfortunately, the access point they are using is actually an attacker's laptop and they're intercepting the original HTTP request and redirecting your employee to a clone of your payroll system instead of the real thing, exposing your employees' personally identifiable information (PII).
If your payroll system uses HSTS and your employee has visited it once using HTTPS, then their browser will know to only use HTTPS, preventing this type of man-in-the-middle attack.
What are the limitations of HSTS?
A key limitation of using HSTS is that a user that cannot connect through HTTPS will be unable to use the site.
Additionally, as the HSTS Policy is communicated in a response header, it requires the user agent to first visit the website to learn that it uses HSTS.
This means the initial request remains unprotected from active attacks if it uses an insecure protocol such as plain HTTP or if the URI for the initial request was obtained over an insecure channel.
This will also apply to the first request after the activity period specified in the HSTS max-age (sites generally set a period of several days or months depending on user activity and behavior).
There is widespread browser support for HSTS including Google Chrome, Mozilla Firefox, Internet Explorer, Microsoft Edge, Opera, and Safari address this limitation by preloading HSTS Policies from the HSTS Preload list. The HSTS list contains known websites that support HSTS and are distributed with the browser so it uses HTTPS for the initial request for the listed websites.
As this approach can never scale across the entire Web, there have been discussions to be able to declare HSTS in DNS records and be access them securely via DNSSEC, which could ensure validity.
Additionally, HSTS is ineffective against typosquatting domains, DNS-based attacks and man-in-the-middle attacks that serve traffic from an artificial domain that is not on the HSTS Preload list.
And as HSTS relies on TLS itself, it also relies on the security of TLS.
Read RFC 6797 for a deeper discussion of the overall HSTS security considerations.
What are HSTS deployment best practices?
If your site is committed to HTTPS and you want to preload HSTS, you must:
- Serve a valid certificate
- Redirect HTTP to HTTPS on the same host, if you are listening on port 80
- Serve all subdomains over HTTPS
- You must support HTTPS for the www subdomain if a DNS record for that subdomain exists
- Serve can HSTS header on the base domain for HTTPS requests:
- The max-age must be at least 31536000 seconds (1 year)must be at least
- The includeSubDomains directive must be specified
- The preload directive must be specified
- If you are serving an additional redirect from your HTTPS site, that redirect must still have the HSTS header (rather than the web page it redirects to)
We recommend you start with the following steps:
- Examine all subdomains and nested subdomains of your site and ensure they work over HTTPS.
- Enable HSTS for all HTTPS responses and ramp up the max-age in stages. During each stage, check for broken pages, monitor your site's metrics, and fix any problems that come up and then wait the full-marx age of the stage before you move on. You can use the following header values:
- 5 minutes - Strict-Transport-Security: max-age=300; includeSubDomains
- 1 week - Strict-Transport-Security: max-age=604800; includeSubDomains
- 1 month - Strict-Transport-Security: max-age=2592000; includeSubDomains
- Once you are confident there are no issues, increase the max-age to 2 years and submit your site to the HSTS Preload list with the preload directive:
- 2 years - Strict-Transport-Security: max-age=63072000; includeSubDomains; preload
- You can add your website to the HSTS Preload list via https://hstspreload.org/.
This is what the Chromium Projects want to see in a preload submission.
What is the history of HSTS?
The HSTS specification (RFC 6797) is based on original work by Collin Jackson and Adam Barth in a paper called ForcedHTTPS: Protecting High-Security Web Sites from Network Attacks which was published in 2008.
On the 18 September 2009, PayPal, Jackon, and Barth published an updated version of the protocol outline in the original paper, available here.
This led to the last "community version" of the then-named "STS" specification published on 18 December 2009, with revisions based on community feedback.
Finally, on 19 November 2012 the HSTS specification (RFC 6797) was published after being approved on 2 October 2012 by the Internet Engineering Steering Group (IESG), a body composed of the Internet Engineering Task Force (IETF) chair and area directors.
What browsers support HSTS?
- Chromium and Google Chrome since version 18.104.22.168
- Firefox since version 4; with Firefox 17, Mozilla integrates a list of websites supporting HSTS
- Opera since version 12
- Safari since OS X Mavericks version 10.9
- Internet Explorer 11 on Windows 8.1 and Windows 7 when KB 3058515 is installed
- Microsoft Edge and Internet Explorer 11 on Windows 10
- BlackBerry 10 Browser and WebView since BlackBerry OS 10.3.3
How UpGuard can check your and your vendors' security controls including HSTS
Companies like Intercontinental Exchange, Taylor Fry, The New York Stock Exchange, IAG, First State Super, Akamai, Morningstar and NASA use UpGuard's security ratings to protect their data, prevent data breaches and assess their security posture.
UpGuard Vendor Risk can minimize the amount of time your organization spends assessing related and third-party information security controls by automating vendor questionnaires and providing vendor questionnaire templates.
We can help you continuously monitor your vendors' external security controls and provide an unbiased security rating.
We base our ratings on the analysis of 70+ vectors including:
- Susceptibility to man-in-the-middle attacks
- Insecure SSL/TLS certificates
- SPF, DKIM and DMARC settings
- HTTP Strict Transport Security (HSTS)
- Email spoofing and phishing risk
- Malware susceptibility
- Unnecessary open administration, database, app, email and file sharing ports
- Exposure to known data breaches and data leaks
- Vulnerable software
- HTTP accessibility
- Secure cookie configuration
- Results of intelligent security questionnaires
We can also help you instantly benchmark your current and potential vendors against their industry, so you can see how they stack up.
For the assessment of your own information security controls, UpGuard BreachSight can monitor your organization for 70+ security controls providing a simple, easy-to-understand cyber security ratings and automatically detect leaked credentials and data exposures in S3 buckets, Rsync servers, GitHub repos and more.
You can read more about what our customers are saying on Gartner reviews.
If you'd like to see your organization's security rating, click here to request your free Cyber Security Rating.