How to Choose a Sandbox

Grab a shovel and start digging through the details

By Mathias Wilder, Area Director and General Manager/EMEA Central, Zscaler

Businesses have become painfully aware that conventional approaches — virus signature scanning and URL filtering — are no longer sufficient in the fight against cyberthreats. This is in part because malware is constantly changing, generating new signatures with a frequency that far outpaces the updates of signature detection systems. In addition, malware today tends to be targeted to specific sectors, companies, or even individual members of a management team, and such targeted attacks are difficult to spot. It has become necessary to use state-of-the-art technology based on behavioral analysis, also known as the sandbox. This blog examines how a sandbox can increase security and it looks at what to consider when choosing a sandbox solution.

The sandbox as a playground against malware
Zero-day ransomware and new malware strains are spreading at a frightening pace. Due to the dynamic nature of the attacks, it is no longer possible to develop a signature for each new variant. In addition, signatures tend to be available only after malware has reached a critical mass — in other words, after an outbreak has occurred. As malware changes its face all the time, the code is likely to change before a new signature for any given type of malware can be developed, and the game starts from scratch. How can we protect ourselves against such polymorphous threats?

There is another trend that should influence your decision about the level of protection you need: malware targeted at individuals. It is designed to work covertly, making smart use of social engineering mechanisms that are difficult to identify as fake. It only take a moment for a targeted attack to drop the harmful payload — and the amount of time between system infection and access to information is getting shorter all the time.

What is needed is a quick remedy that does not rely on signatures alone. To detect today’s amorphous, malicious code, complex behavioural analysis is necessary, which in turn requires new security systems. The purpose of a sandbox is to analyse suspicious files in a protected environment before they can reach the user. The sandbox provides a safe space, where the code can be run without doing any harm to the user’s system.

The right choice to improve security
Today’s market appears crowded with providers offering various solutions. Some of them include virtualization technology (where an attack is triggered through what appears to be virtual system) or a simulated hardware solution (where the malware is offered a PC), through to solutions in which the entire network is mapped in the sandbox. However, malware developers have been hard at work, too, and a well-coded package can recognize whether a person is sitting in front of the PC, it can detect if it’s in a virtual environment in which case it can alter its behavior, and it can undermine the sandboxing measures by delaying activation of the malicious code after infection. So, what should companies look for when they want to enhance their security posture through behavioral analysis?

What to look for in a sandbox

  • The solution should cover all users and their devices, regardless of their location. Buyers should check whether mobile users are also covered by a solution.
  • The solution should work inline and not in a TAP mode. This is the only way one can identify threats and block them directly without having to create new rules through third-party devices such as firewalls.
  • First-file sandboxing is crucial to prevent an initial infection without an existing detection pattern.
  • It should include a patient-zero identification capability to detect an infection affecting a single user.
  • Smart malware often hides behind SSL traffic, so a sandbox solution should be able to examine SSL traffic. With this capability, it is also important to look at performance, because SSL scanning drains a system’s resources. With respect to traditional appliances, a multitude of new hardware is often required to enable SSL scanning — up to eight times more hardware, depending on the manufacturer.
  • In the case of a cloud sandbox, it should comply with relevant laws and regulations, such as the Federal Data Protection Act in Germany. It is important to ensure that the sandboxing is done within the EU, ideally in Germany. The strict German data protection regulations also benefit customers from other EU countries.
  • A sandbox is not a universal remedy, so it should, as an intelligent solution, be able to work with other security modules. For example, it is important to be able to stop the outbound traffic to a command-and-control (C&C) centre in the case of an infection. In turn, it should be possible to turn off the infected computer by tracing back the C&C communication.

Putting it all together
All these criteria can be covered by an efficient and highly integrated security platform, rather than individual hardware components (“point” appliances). One advantage of such a model is that you get almost instantly correlated logs from across the security modules on the platform without any manual interaction. If a sandbox is part of the platform, the interplay of various protection technologies through the automated correlation of data ensures faster and significantly higher protection. This is because it is no longer necessary to feed the SIEM system manually with logs from different manufacturers.

Platform models do not lose any information as they allow all security tools — such as proxy, URL filters, antivirus, APT protection, and other technologies — to communicate with one another. It eliminates the time-consuming evaluation of alerts, as the platform blocks unwanted data extraction automatically. A cloud-based sandbox together with a security platform is, therefore, an effective solution. It complements an existing security solution by adding behavioral analysis components to detect previously unknown malware and strengthens the overall security posture — without increasing operating costs.

Self-Driving Information Security

By Jim Reavis, Co-founder and CEO, Cloud Security Alliance

The prospects of autonomous self-driving vehicles becoming a pervasive presence on our roadways seems more likely everyday. From the big automakers to Tesla to Google to Uber, a wide range of companies are investing a tremendous amount of money to create a world without carbon-based drivers. The motivation for a big payday abounds, but the hope is that this will be a huge boon to vehicle safety, and I believe it ultimately will be. As we have learned at hacker conferences, there are a lot of security concerns about self-driving cars that we need to solve, but that is not what I want to talk about here.

What I would like to do here is steal the term from the automotive industry and apply “Self Driving” to Information Security. What is Self-Driving Information Security? For me, this is an initiative to apply the ever growing power of computing to solve complex and fast changing information security problems dynamically and without human intervention. Do I believe we can eliminate humans from the information security industry? No, I don’t believe that is possible or desirable, and it certainly would make BlackHat a lot less fun. However, I think we need to rapidly take steps to push the envelope on where we can take the person out of the loop, simply because we are not going to have enough humans to go around and insert into every potential security problem space. In a world where we will soon have thousands of Internet connected devices for every person on Earth, it’s highly unlikely we will have enough information security professionals to go around to solve all of the resultant problems.

Automation is a very old idea that is present in every industry. In information technology, we seek to automate every repetitive task we can. But like in other industries, the explosion in compute power is causing us to explore automating ever more sophisticated tasks. It is no longer just assembly line robots, but advances in computing are taking on white collar jobs and in many cases doing a great job. Computers are diagnosing diseases more accurately than doctors. Computers are doing journalism and even taking on the legal profession.

Are you a skeptic in regards to computer encroachment on sophisticated and complex professions? One of the most seminal moments in computing history that impacted me was the chess contest between Garry Kasparov and IBM Deep Blue. Personally, I was rooting for the human until the bitter end. When Deep Blue ultimately defeated the world’s greatest chessmaster, I was in mourning for days. That was 20 years ago.

To be clear, Self-Driving Information Security will not be bereft of humans. Humans are the biggest part of information security today by any measure – clearly by the budgetary metric. I think we will continue to grow the overall number of people employed in the profession for the foreseeable future. The unpredictability of information security and its adversarial, logic-defying nature will require humans. But Self-Driving Information Security will gobble up the jobs we are doing today, and I am not quite sure what jobs we will be doing in the future. What I do know is, if we do not implement Self-Driving Information Security, we are going to drown in information and incidents.

What are some of the building blocks of Self-Driving Information Security? It is actually many things we are working on today, they just need to gain maturity:

DevSecOps. This idea of merging DevOps with Security Operations, enabled by cloud, is gaining in popularity with very diverse security teams. The ability to tear down and instantiate new computing systems, using “serverless” capabilities and applying some imagination is leading to automation of security process that can seem like magic to an old security guy like me.

Autonomics. The ability for computers be self-managing, self-healing, self-optimizing – self-EVERYTHING is important. A big part of how the Internet works today is through some levels of hierarchy and “command and control” systems. Clearly this model is going to break. I think about the apartment of the future with thousands of computers. Then I think about the bad guy that attacks the upstream link or servers. Or perhaps malware is injected into one of the apartment’s devices. In both cases, the nodes must not only be resilient and independent, but may need to collaborate and attack the infected device.

Blockchain. The distributed, immutable ledger technology that underpins Bitcoin is a favorite of VCs and the finance industry. I believe we are going to find a lot of applications for Bitcoin in information security. An authoritative, tamper-proof log of transactions which can be either public or private has fascinating implications. We can record any change in a very granular manner. I think about IT audit and having a record of all security control implementations, it can really change how that job is done.

Analytics. Data Science. The answers are in the data. If the data sets are large enough, if the quality is good enough and if the algorithms are well designed and speedy, we will find the security answers we are looking for. I believe our massive and inexpensive compute infrastructure is going to excel in finding the right answer to a new security problem

Artificial Intelligence. AI is certainly controversial, even trying to define it can cause fights. Many are terrified by AI and its potential threat to mankind. Some security solutions claim to use AI, others say that the current products are really employing machine learning. Closely related to analytics, having access to quality data is going to enable AI to make security decisions and take action before a human can blink.

In addition to all of these areas of focus, it is safe to assume that computing is going to get faster, cheaper and bigger at an ever-increasing pace. Quantum computing may be years away, but there are already serious efforts in government and industry to make a massive leap in computational speed. It’s also safe to assume that the bad guys will want to harness or exploit all of these trends for themselves.

The building blocks above will soon be assembled together into Self-Driving Information Security. It will be quite necessary for this to happen to manage our rapidly increasing compute universe. The jobs we know today will go away. I am convinced new jobs will replace them in greater numbers, but it may be messy. The paradox of automation is that humans will operate in a world with more layers of complex technical abstraction. We aren’t as intimately involved, but when we are needed, it is for more critical reasons.

At Cloud Security Alliance, we think it is important to be considering these trends now to be true to our mantra of “solving tomorrow’s problems today”. That’s why we have research in all of these areas happening in 2017. As always, our research is your research and we encourage you to join us.


There May Be a Shark Circling Your Data

By Jacob Serpa, Product Marketing Manager, Bitglass

In today’s business environment, cybersecurity remains a topic of great importance. As more companies migrate to the cloud, security concerns continue to evolve. While BYOD (bring your own device) affords employees more flexibility as they work from a multitude of devices, it also exposes data to nefarious parties in new ways. In the face of increasingly sophisticated cyber attacks, companies must learn and adapt or suffer the consequences.

In its latest cybersecurity report, “Threats Below the Surface,” Bitglass discusses the results of its survey of over 3,000 IT professionals. With the help of the CyberEdge Group and the Information Security Community, Bitglass was able to uncover the threats, priorities, and capabilities seen as most relevant by these professionals. The fact that the last year has seen 87% of organizations become victims of cyber attacks (and that a third of those organizations were hacked over five times) lends credence to cybersecurity concerns.

Despite the importance of maintaining visibility into data usage, relatively few firms are doing it well. While over 60% of companies monitor their desktops, laptops, and networks for security threats, the percentage drops to 36% for mobile devices and 24% for SaaS and IaaS applications. As organizations (inevitably) adopt BYOD and public cloud apps for increased productivity, they should proactively monitor for the corresponding security risks. However, when the survey respondents were asked about their firms’ current security postures, they indicated that they were primarily concerned about vulnerability with respect to mobile devices. Other prominent concerns included malware, privacy, and data leakage.

While most companies plan to increase their security budgets for next year, they should already be taking steps to ensure cybersecurity systems that consider contemporary tools like the cloud and BYOD. In particular, firms should be utilizing end-to-end solutions that secure data on devices, in transit, and at-rest in the cloud, while addressing concerns about topics like privacy.

More and more, conscientious companies are turning to CASBs (Cloud Access Security Brokers) and UEBA (user and entity behavior analytics) for modern-day cybersecurity. CASBs allow for discovering shadow IT apps, ensuring regulatory compliance in the cloud, preventing unwanted data disclosures, and more. With UEBA, a core component of CASBs, enterprises can detect account hijacking, data exfiltration, and other threats. CASBs and UEBA give companies a great deal of visibility and control over their data – a huge help in keeping an eye on the threats below the surface.

The Cure for Infectious Malware

By Chantelle Patel, Marketing Manager, Bitglass

Organizations have seen rapid growth in cloud adoption over the last few years which in turn have introduced new threats and increased the risk of data leakage. Among the most prominent threats are malware and ransomware – long a problem on endpoints. With the advent of public cloud apps, interconnected and widely used, malware and ransomware have the potential to touch more data than ever before.

Unfortunately, despite the risk to data in the cloud, few providers offer any malware protection whatsoever. Those that do offer limited signature-based threat protection, based on solutions from IPS/IDS vendors, can only identify known threats. The most dangerous threats are not these known pieces of malware, but the unknown, zero-day threats that can go undetected, resulting in weeks or months of data exfiltration unbeknownst to IT.

Some solutions offer threat protection that is reactive rather than proactive, and what little proactive protection they provide is ineffective when end-users need instant access to data in the cloud or expect instant upload of a file. This gets at a critical difference between traditional signature-based malware and next-generation AI-based malware. Traditional tools rely on dynamic analysis, executing a file in a sandbox before taking action. Next-generation tools from companies like Cylance leverage static analysis, basing a risk decision on hundreds of characteristics associated with a file.

Once malware makes its way into a cloud app, there’s little an organization can do to stop its spread. These malicious files are often downloaded to endpoints, make their way to connected apps, and are shared across the organization. The only way to protect against these threats is to prevent their spread.

With Advanced Threat Protection (ATP), a core component of any complete Cloud Access Security Broker (CASB) solution, organizations can protect the cloud from malware before it hits the app, assess the risk of any one file, and stop malicious attacks in their tracks.

Why You Need a CASB for GDPR Compliance

By Rich Campagna, Senior Vice President/Products & Marketing, Bitglass

With enforcement of the EU’s General Data Protection Regulation (GDPR) is just over a year away in May, 2018, your planning efforts should already be well underway. Adoption of cloud applications across the EU continues at a rapid clip, and the global nature of leading cloud applications means that protecting personal data and achieving data residency can be difficult to achieve.

With mandatory breach notifications and very steep fines (4% of annual revenues), the cost of non-compliance is high. On the other hand, it’s nearly impossible to stop the move to cloud in most organizations, so that’s not an option either. Fortunately, you still have time to arm your organization with the key to combining cloud adoption and GDPR compliance: cloud access security broker (CASB). Let’s take a look at some of the areas where a CASB can help:

  • Identifying personal data – the EU GDPR is primarily concerned with the protection of any data that can be used to identify a person (name, address, email, driver’s license number, and much, much more). The first thing that you need to do in order to protect that data is to identify where it is. CASBs can scan across both data-in-transit and data-at-rest for a wide range of cloud-delivered apps (SaaS, IaaS, and custom applications). Any CASB you choose should have a library of pre-built identifiers that can be used to scan for names, phone numbers, addresses, national identity and driver’s license numbers, health record information, bank account numbers, and more.
  • Controlling the flow of personal data – Once you’ve identified where sensitive data resides, you want to control where it can go. CASBs include a range of policy options that allow you to do things like geofence personal information, control access from unmanaged/unprotected devices, control external sharing, and encrypt data upon download. All of these options can help mitigate the risk of non-compliance.
  • Maintaining data residency and sovereignty – Major cloud applications often have global architectures which makes it difficult, it not impossible, to keep data within a given country or region. Fortunately, the GDPR allows for the use of encryption to meet GDPR requirements, if the cloud provider transfers data outside of the EU. Seek out a CASB that offers the killer app for GDPR – full-strength cloud encryption – across both unstructured (file) and structured (field) data.
    • Word of caution: some cloud application vendors offer their own “built-in” or “platform” encryption. With these schemes, the cloud provider has access to the keys and, therefore, the data as well. This is a GDPR gray area and may leave you, the data controller, on the hook for those hefty fines and mandatory notifications.
  • Monitor Risky Activity – A CASB can give you visibility into everything that’s happening with your users and your data across protected cloud applications. User Behavior Analytics and alerting capabilities let you know when risky activity is happening. This might mean reporting on indicators of breach, credential compromise, personal data access from outside the EU, or more. This critical visibility will allow you to identify and stop activities that might otherwise leave you staring down a fine of 4% of revenues (and a corresponding loss of your job).
  • Identify Shadow IT – simply put, GDPR and Shadow IT are a volatile and risky mix. There’s simply no feasible way to get the controls and visibility needed over applications that your organization has no ability to control. A CASB can give you much needed visibility into Shadow IT applications, and their corresponding risk, but your only option when faced with GDPR is get out of the shadows – either sanction and protect shadow IT through a CASB, or block unsanctioned apps altogether.

These CASB controls can really jumpstart a successful GDPR program across your organization, leaving you free to consider some of the many other GDPR-related controls and policies you’ll need to put in place over the next 12 months, including appointing a Data Protection Officer, figuring out how to implement “right to be forgotten,” and reevaluating licensing terms and data ownership across your many cloud application vendors.