Web Application Firewalls (WAFs) have been in production use for well over a decade, maturing from point solutions primarily blocking SQL injection to mature application security tools. In most mature security product categories, such as anti-virus, there hasn’t been much to talk about, aside from complaining that not much has changed over the last decade. WAFs are different: they have continued to evolve in response to new threats, new deployment models, and a more demanding clientele’s need to solve more complicated problems. From SQL injection to cross-site scripting (XSS), from PCI compliance to DDoS protection, and from cross-site request forgeries (CSRF) to 0-day protection, WAFs have continued add capabilities to address emerging use cases. But WAF’s greatest evolution has taken place in areas undergoing heavy disruption, notably cloud computing and threat analytics.
WAFs are back at the top of our research agenda, because users continue to struggle with managing WAF platforms as threats continue to evolve. The first challenge has been that attacks targeting the application layer require more than simple analysis of individual HTTP requests – they demand systemic analysis of the entire web application session. Detection of typical modern attack vectors including automated bots, content scraping, fraud, and other types of misuse, all require more information and deeper analysis. Second, as the larger IT industry flails to find security talent to manage WAFs, customers struggle to keep existing devices up and running; they have no choice but to emphasize ease of set-up and management during product selection.
So we are updating our WAF research. This brief series will discuss the continuing need for Web Application Firewall technologies, and address the ongoing struggles of organizations to run WAFs. We will also focus on decreasing the time to value for WAF, by updating our recommendations for standing up a WAF for the first time, discussing what it takes to get a basic set of policies up and running, and covering the new capabilities and challenges customers face.
WAF’s Continued Popularity
The reasons WAF emerged in the first place, and still one of the most common reason customers use it, is that no other product really provides protection at the application layer. Cross-site scripting, request forgeries, SQL injection, and many common attacks which specifically target application stacks tend to go undetected. Intrusion Detection Systems (IDS) and general-purpose network firewalls are poorly suited to protecting the application layer, and remain largely ineffective for that use case. In order to detect application misuse and fraud, a security solution must understand the dialogue between application and end user. WAFs were designed for this need, to understand application protocols so they can identify applications under attack. For most organizations, WAF is still the only way get some measure of protection for applications.
For many years sales of WAFs were driven by compliance, specifically a mandate from the Payment Card Industry’s Data Security Standard (PCI-DSS). This standard gave firms the option to either build security into their application (very hard), or protect them with WAF (easier). The validation requirements for WAF deployments are far less rigorous than for secure code development, so most companies opted for WAF. Shocking! You basically plug one in and get a certification. WAF has long been the fastest and most cost-effective way to satisfy Requirement 6 of the PCI-DSS standard, but it turns out there is long-term value as well. Users now realize that leveraging a WAF is both faster and cheaper than fixing bug-ridden legacy applications. The need has morphed from “get compliant fast!” to “secure legacy apps for less!”
The value of WAF is muted by difficulties in deployment and ongoing platform management. A tool cannot provide sustainable value if it cannot be effectively deployed and managed. The last thing organizations need is yet another piece of software sitting on a shelf. Or even worse an out-of-date WAF providing a false sense of security.
Our research highlighted the following issues which contribute to insecure WAF implementations, allowing penetration testers and attackers to easily evade WAF and target applications directly.
- Ineffective Policies: Most firms complain about maintaining WAF policies. Some complaints are about policies falling behind new application features, and policies which fail to keep pace with emerging threats. Equally troubling is a lack of information on which policies are effective, so security professionals are flying blind. Better metrics and analytics are needed to tell users what’s working and how to improve.
- Breaking Apps: Security policies – the rules that determine what a WAF blocks and what passes through to applications – can and do sometimes block legitimate traffic. Web application developers are incentivized to push new code as often as possible. Code changes and new functionality often violate existing policies, so unless someone updates the whitelist of approved application requests for every application change, a WAF will block legitimate requests. Predictably, this pisses off customers and operational folks alike. Firms trying to “ratchet up” security by tightening policies may also break applications, or generate too many false positives for the SoC to handle, leading to legitimate attacks going ignored and unaddressed in a flood of irrelevant alerts.
- Skills Gap: As we all know, application security is non-trivial. The skills to understand spoofing, fraud, non-repudiation, denial of service attacks, and application misuse within the context of an application are rarely all found in any one individual. But they are all needed to be an effective WAF administrator. Many firms – especially those in retail – complain that “we are not in the business of security” – they want to outsource WAF management to someone with the necessary skills. Still others find their WAF in purgatory after their administrator is offered more money, leaving behind no-one who understands the policies. But outsourcing is no panacea – even a third-party service provider needs the configuration to be somewhat stable and burned-in before they can accept managerial responsibility. Without in-house talent for configuration you are hiring professional services teams to get up and running, and then scrambling to find budget for this unplanned expense.
- Cloud Deployments: Your on-premise applications are covered by WAFs and you’re reasonably happy with your team’s technical proficiency, but as your company moves applications into public cloud infrastructure (and they are, whether you know about them or not), you may be unable to migrate your WAF and associated policies to this new and fundamentally different architecture. Your WAF vendor may not provide a suitable substitute for the appliance you run on-premise, or it might not offer the APIs you need to orchestrate WAF deployment in your dynamic cloud environment, where applications and servers come and go much faster than ever before.
Going It Alone
If all those problems sound like a nightmare, don’t worry – you’ll still be able to sleep tonight. The nightmare scenario is actually not using some type of WAF or filter to protect your applications, and instead leaving them wide open to attack. Sure, a few of you have been fortunate enough to build your applications from scratch with security in mind, which is awesome. But those unicorns are rare – the rest of us are relegated to fixing vulnerabilities in existing applications… you know, the ones designed and coded without any input from security.
In most software stacks current and future defects are discovered in the underlying platform, third-party software, and in-house applications. That’s just software. Of course attackers know this, so they probe applications for new vulnerabilities. Legacy platforms will take years, if not decades, to meet modern security requirements. And the cost of these efforts is inevitably many times greater the original investment. In many cases it is simply not economically feasible to fix the application, so some other approach must be used – which is where WAF offers tremendous value.
Our research shows that WAF failures far more often result from operational failure than from fundamental product flaws. Make no mistake – WAFs are not a silver bullet – but a correctly deployed WAF makes it much harder to successfully attack an application, and for an attacker to avoid detection. The effectiveness of WAFs is directly related to the quality of people and processes used to maintain them. The most serious problems with WAF are more issues with management and operational processes than technology. We need a pragmatic process to manage Web Application Firewalls to overcome the issues which plague the technology.
Our next post will offer advice on how to deploy WAFs and look at some recent feature enhancements which address the ease of use and effectiveness issues mentioned above. We will also highlight some Quick Wins for new setups, because any new security technology needs an early win to prove the value of an approach and the effectiveness of a technology.