Wednesday, November 12, 2008

A new type of network appliance

One of my friends sent me this patent to see if it was something that our company would be interested in. After reading the abstract and summary it reinforced my belief that there are several applications in the world where a network appliance fits in to a high availability architecture, but there always needs to be a balance between infrastructure complexity and stability.

If you are interested in what another California inventor has conjured up, read on.

United States Patent Application 20080016570


The method analyzes unauthorized intrusion into a computer network. Access is allowed through one or more open ports to one or more virtualized decoy operating systems running on a hypervisor operating system hosted on a decoy network device. This may be done by opening a port on one of the virtualized decoy operating systems. A network attack on the virtualized operating system is then intercepted by an introspection module running on the hypervisor operating system. The attack-identifying information is communicated through a private network interface channel and stored on a database server as forensic data. A signature-generation engine uses this forensic data to generate a signature of the attack. An intrusion prevention system then uses the attack signature to identify and prevent subsequent attacks. A web-based visualization interface facilitates configuration of the system and analysis of (and response to) forensic data generated by the introspection module and the signature generation engine, as well as that stored in the processing module's relational databases.


One or more embodiments of the invention are directed to an improved method and system for protecting computer networks. In one embodiment, the invention comprises a modular decoy network appliance, which runs fully functional operating systems on client hardware modules. The modular arrangement comprises front-end fully functional operating system modules and a separate processing back-end module.

The front-end presents a standard fully functional operating system, such as Windows® or a flavor of Linux®, or Sun Microsystems Solaris® that returns a standard operating system fingerprint when it is scanned by tools that attackers typically use to identify vulnerable systems. The attacker is thus lured into accessing the identified operating system and running custom or known exploits on that system.

The front-end module includes a sentinel kernel driver (or a more generalized executable module) that is hidden from system scanners as it is removed from kernel module listings or registry in Windows. Thus, the kernel does not indicate the sentinel kernel driver is running. The sentinel kernel driver monitors connections to the operating system as well as activity on the operating system and activity on services running on the operating system. When an attacker connects to a port, the sentinel kernel driver captures the data coming through the socket. Generally all relevant data coming through the socket is captured. In most cases this means whatever data is received as part of an incoming attack is captured by the sentinel driver. Captured data is sent as a slew of common UDP packets to the back end processing module over the fabric network connection separate from the vulnerable front-end modules. In this manner, there is no way for the intruder to know that his or her communications with the operating system are being analyzed.

The captured data, which contains the attack-identifying information, is sent to the back-end or processing module though the backplane fabric of the appliance using Layer 2 Ethernet communication protocol. The processing module is separate and independent from the client operating system modules and communicates the processed information to security administrators through a network port connected to the private and secure VLAN. Unbeknownst to the intruder, the exploit is thus captured, transferred and analyzed.

With the received data, the processing module generates a report of the attack. The report consists of user-friendly information that paints a picture of the attack for a system administrator. This may include information on which sockets were accessed, what happened at a particular socket, the key strokes entered or bytes transferred to the port, what files were transferred, registry changes, how the attack was run, what happened on the primary network, on its servers or how the network services were affected. The report may also include information on the location of the attacker or the attacker's service provider. Graphical representations of key information and interactive mapping of the attack locales by region or country may be utilized in one or more embodiments of the invention.

The processing module is used to generate an attack signature by analyzing all the data passed through the socket. The signature is generated by analyzing the attack payload including the keystrokes or transferred bytes and any files uploaded to the client operating system of an ASCII or binary nature. The files uploaded are assumed to be of a malicious nature created to deliver a malicious payload in the form of a compiled program or an interpreted script. In the event that no malicious files are uploaded to the operating system, the signature generation engine analyzes all the keystrokes or bytes delivered through the socket and creates a pattern signature which when applied to an IDS or IPS system, enables the IDS or IPS systems to detect the attack if repeated on production systems. Once generated, the attack signatures can be viewed by a system administrator to determine the appropriate course of action. The system administrator can instruct the signature to be uploaded to the intrusion detection system (IDS) or intrusion prevention system (IPS) for the protected network where it is added to the IDS's or IPS's library of signatures to protect production systems. In one or more embodiments of the invention, the signature may be uploaded or saved in a third party system that maintains all known exploits. In this manner, other systems may be notified through secure channels of an impending threat. For example, by transferring the signature to a centralized server that communicates with multiple installations, the intruder may be thwarted before attacking other systems in other companies.

No comments: