In the recently published 2015 SonicWall Security Threat Report, one of the observations on the evolution of attacks on POS systems is the rise in popularity of malware that uses memory scraping to steal sensitive data. No matter how many layers of encryption are applied to sensitive payment data and how carefully this encryption is deployed, at some point the primary account number and other sensitive information must exist in an unencrypted form in order to be useful. The moment that payment data is decrypted for processing, it ends up in the memory of the POS machine, creating a perfect window of opportunity for an attacker to snag this data. Advanced malware can use multiple techniques to access and scan contents of this temporary storage and look for patterns that resemble raw payment data. This data can then be used, for example, to clone cards for fraudulent purchases. This is exactly what happened in some of the high profile retail breaches of 2013 and 2014.
The ultimate goal of RAM scraping malware is exfiltration of the unencrypted data stolen from memory of the infected machine. Therefore, this malware will be very well hidden and it will attempt to remain as invisible as possible in order to access as much data as possible. Mitigating the risks of being hit with such malware falls into two categories: Pre-infection best practices to avoid infection and post-infection best practices to detect and control the attack.
Pre-infection best practices
Protecting yourself from new advanced attacks must always be done on top of executing on the basics which serve to reduce the risk of getting critical systems such as POS systems infected by any malware.
- Keep the OS and applications on POS systems fully patched. Most patches are security related, so ignoring them only opens up a larger window of opportunity for attackers.
- Firewall off the POS network from the rest of the network with strong (i.e. bare minimum access) access policies as well as with Intrusion Prevention and Anti-Malware.
- Use strong, non-default and not shared, passwords.
- Deploy and enforce endpoint anti-virus as a last measure of defense.
- Encrypt traffic VPN tunnels.
- Enable protection against MAC spoofing within the POS network and for critical systems with which the POS terminals communicate.
- Lock down remote access to pin-point level of access. Do not allow full L3 tunnels into sensitive networks and use remote access tools that allow verification of remote host integrity before granting access.
Post-infection best practices
A good to approach in evaluating your network security stance is to assume that you will be infected at some point in the future and design processes to allow you to detect and control the infection. In the context of memory scraping malware, the ultimate observable behavior will be communication with non-trusted hosts on the internet. It may not be immediate and it may not be in bulk, as the attacker may want to put time between the act of infection and the act of data theft. However, sooner or later, the attacker will need to get the stolen data from the POS systems into his or her possession. This may happen naively via direct communication, or via more sophisticated methods such as using another compromised system outside the POS network, but with a connection to the POS network, as a gateway. That system may reside in a network that is less strictly observed than the POS network on which may not raise alarms at communication with random servers on the internet.
There are several key technologies that can help you detect or neutralize this data exfiltration:
- Don’t allow direct communication with the internet from the POS network. This will lock down allowable communications and will block and detect naÃ¯ve approaches at data exfiltration. For processing purposes, payment data can be sent via an encrypted tunnel to another trusted server(s) on the network (outside the POS network) and then via another encrypted tunnel to the processing server. Communication between these systems should be whitelisted by the firewall via ACLs, with all other traffic (besides perhaps management and updates) blacklisted.
- Deploy Geo-IP and Botnet filtering detection on all networks. Lock down communication from sensitive systems only to locales that they need to communicate with (if your processor is in the US, why would your POS data need to have access to and from Europe, Asia, LATAM, etc.?)
- Configure DLP and SSL Decryption to detect Credit Card type data leaving the network in plaintext or inside of SSL tunnels to internet hosts that are unknown. In other words, only allow such data to flow to CC processing servers known to you. Communication of such data to any other system on the internet should be intercepted, logged and investigated. Deny any SSL communication from sensitive networks that does not lend itself to inspection by not accepting your NGFW SSL inspection certificate.
Firewalls occupy an extremely valuable piece of real estate on any network since all Internet bound traffic must go through them. When properly deployed, next-generation firewalls play an important role in reducing the risk of advanced malware infection and data theft in POS networks. To find out more about the capabilities of state of the art NGFWs from SonicWall, read the eBook “Types of Cyber-Attacks and How to Prevent Them.” Follow me on Twitter: @threadstate.