This is the third article in a three-part series about considering best practices for deploying IP security systems. The discussion is about the wider scope of best practices for deploying physical security technology on enterprise networks. Such practices are needed because many security devices and systems were designed on the assumption that the equipment would be deployed on a completely independent security network rather than in an enterprise network environment.
Thus the authors have formed the Bp.IP Initiative, to advance best practices for deploying IP based security systems in enterprise network environments, including practices that compensate or work around the network-related shortcomings of security products currently available.
An enterprise network environment is a managed network environment. In a managed network:
For example, in a well-managed enterprise network a port on a network switch will be shut down automatically if an unacceptable network traffic condition occurs. There is usually a critical data backup scheme, whereby a standard backup solution is put into place and utilized as appropriate for various classes of data. Backup and emergency power provisions keep critical portions of the network operational (along with critical information systems) during power outages. Redundant network paths are designed and managed so that loss of a single pathway triggers a rerouting of affected network traffic.
Readers with a technical interest in Syslog should see the overview available at: http://www.ciscopress.com/articles/article.asp?p=426638
In March of 2009, the syslog protocol was updated (as RFC 5424) to provide a message format that allows vendor-specific extensions to be provided in a structured way. A number of network cameras from Axis, Cisco and others support syslog messaging.
Requirements for Security Systems and Devices
In a managed network environment, security systems and devices that connect to, or operate on, the network must not generate prohibited traffic or in other ways act outside the acceptable parameters defined for network computers and devices.
Additionally, the computer and network security of the security systems and devices must meet or exceed the minimum standards of the network. This prevents situations that have plagued some enterprise IT departments. For example, DVRs that buyers thought were “recording devices” turned out to be Windows-based computers with no antivirus software and unpatched older versions of the operating system, making excellent sites for virus infections that then spread from the DVRs throughout the enterprise network, disabling the video systems in the process. Vulnerable systems and devices can bring havoc to a network when compromised. (Download a study from Cisco about its reaction to such an experience with DVRs: http://www.bpforip.com/downloads/Cisco_IT_Case_Study_CCTV.pdf.)
Finally, security systems and devices should support network management by incorporating appropriate network management features, such as syslog and simple network management protocol (SNMP), to name just two.
IT evaluators are personnel assigned the responsibility of evaluating systems and devices for approval to be placed on the enterprise network. Since most currently available security systems and devices were not specifically designed to be installed and operated in a managed network environment, many products would fail such an evaluation unless an appropriate set of deployment recommendations were also developed and provided.
Following the ISC West 2010 event, the authors assessed a number of access control and video surveillance products (some equipment provided directly by vendors, and some currently deployed in field installations). The purpose was to establish what would be needed – be it product configuration tuning, compensating controls, or workaround procedures – to enable the products to be soundly deployed in an enterprise network environment. The assessment criteria were selected to cover a reasonable but broad sample of enterprise-class networking features that can be judiciously applied to physical security solutions. General recommendations were developed that can be fine-tuned based upon the specifics of any particular technology being considered for deployment.
The basic assessment criteria the author’s applied are listed below. Note that some of the assessment criteria require thinking about the use to which the security system or device will be put, as well as consideration of the network environment in which the products will be deployed.
Basic findings and recommendations are provided in summary form below, in the following categories:
In the space of this article, it is not feasible to provide a complete examination of each type of finding and related sound deployment practice. This article is intended to introduce evaluation criteria commonly used in evaluating networked systems and devices, and provide some rationale for applying them to security system computers and devices.
A key recommendation for all networked security systems is to put the computers and devices on a dedicated private subnetwork, which provides logical separation of network traffic running on the same physical cable (see the Wikipedia topic: Subnetwork). This improves network performance and helps make sure the systems and devices have a stable environment to operate in without interference from any other network traffic sources.
Security systems, like any other server-based software systems, have backup and restore requirements, and the typical infrastructure requirements for power, heat dissipation (heat is what shortens the life of many DVR/NVR hard drives), and server software maintenance. The systematic application of maintenance upgrades is a practice taken for granted in the IT networking world – in the physical security world it can require extra effort to obtain version information and to learn about what mechanisms may exist to manage the application of updates to servers and devices in a large system.
The broad category of “video servers” includes the VMS (Video Management System – which is software on a server), DVR (Digital Video Recorder), NVR (Network Video Recorder), Hybrid NVRs (NVRs that support both analog and network cameras) as well as video analytics software.
Some systems are Windows-based machines, as mentioned earlier in this article, and have specific vulnerabilities, and some are proprietary embedded systems that do not have the vulnerabilities of Microsoft Windows®. Many brand name DVRs and NVRs, although they function as servers, use desktop versions of the operating system (Windows 98, Windows 2000 Professional, and Windows XP). Many (not all) are not capable of running anti-virus and other security software, as their processors are not sufficiently powerful to handle both video recording and any other significant task. Video servers inherit the vulnerabilities of their operating system. Thus they need to be kept current with operating system updates (patches) for the sake of system and network security. Some operating system video components, such as Windows DirectX, should have strict update hygiene applied for cyber-security reasons as well as performance reasons.
Leading VMS software provides the opportunity to use network or operating system features to get the level of strong authentication that should be established, for example, for remote login over the network. Thus some applications should be set up using the vendor-supplied features that take advantage of the underlying Windows or network login mechanisms. Where the network infrastructure provides one-time password support, it should be utilized.
For a high level of physical and logical authentication (such as for protecting server rooms and equipment closets), biometric authentication can be utilized to fortify access to specific doors, systems and applications. The plusID® device from Privaris (www.privaris.com) is a key fob device that provides a way to add personal biometric authentication to any access control system, as it emulates up to four proximity cards and smart cards and works without changing existing access control systems and card readers. It can be used (via USB connection) to enforce biometric logon to servers, workstations and laptops. It also supports RSA SecureID® one-time password and is interoperable with existing SecureID deployments.
Many network video cameras used clear text (unencrypted) transmissions by default. Some cameras have difficulty running an encrypted connection at high resolutions and frame rates. In this situation using a dedicated management subnetwork or Virtual LAN (VLAN) to compensate is recommended. This is to provide the appropriate privacy to this sensitive information. Most network cameras (and network encoders for use with analog cameras) support HTTPS (Hypertext Transfer Protocol Secure), which is a combination of HTTP and TLS/SSL. (Transport Layer Security is the successor to Secure Sockets Layer—for more information see the Wikipedia entry for Transport Layer Security.)
While monitoring for video loss is an important video software feature, network cameras should also have network monitoring applied. Some network cameras can have their passwords and IP addresses reset to the default by pressing the reset buttons. The default IP addresses and the camera reset procedures are published on the Web for most major brands (installer manuals, user guides, online FAQ, support forums, etc.).
Simply pressing a reset button to cause an IP address reset may take the camera out of the subnet that it is on. It takes knowledge of the default IP address to find it again (as it is has moved to a different logical network segment). Where a fixed IP addressing scheme is in use, resetting multiple cameras to the same default IP address can take all but one off the network due to the IP address conflict.
In a well-managed network, there is logged network activity resulting from a network camera reset. Log messages including “device offline” and “IP address conflict” would be reported by the network management software. The nature of the messages can provide indications of the type and location of trouble.
Access control servers can be vulnerable to cyber attacks. See the May Convergence Q&A Column titled “The Security World Has Changed” for a description and reference information on the access control system attack that was documented at CarolinaCon, an annual hacker’s conference in North Carolina. Vulnerabilities are specific to each access control server, but some vulnerabilities are not brand-specific and apply to many brands.
For example, a number of access control systems require that one specific password be used for the SQL Server database; all systems use the same database password. Sometimes the passwords are contained in the installer or user manuals downloadable from the vendor’s website. This makes it possible for anyone with basic database knowledge – and any brand’s password – to access any access control databases of that brand. Such access can often be accomplished across the network – physical access to the server is not necessary. Where database passwords can be changed, many integrators leave the default password in place or use the same password for all of their customers, to keep things easy for the service technicians. They should be changed when technicians leave employment, but that doesn’t happen in these situations. Integrators should be asked about their approach to password management, and the response considered during integrator selection.
Access control equipment (control panels, system controllers, door controllers, IP readers – various items that provide the hardware-based distributed intelligence) can be vulnerable to outside network interactions, including intentional attacks. Additionally, some systems introduce proprietary “plug and play” traffic or device discovery traffic that can appear to be an attack to some types of network monitoring software. Some systems provide DNS server functionality (to enable automatic IP addressing), and there must only be one such DNS server on any network segment. Utilizing a subnetwork will isolate such network traffic from other systems and devices on the enterprise network.
High caliber consultants and systems integrators attend to many of these issues, but often do so without educating clients and customers about the protective measures, and without documenting them sufficiently. Many integrators give little attention to these issues.
Security system designs should include a computer and network security plan, which addresses sound deployment practices appropriate for the specific types and brands of software and hardware being deployed. Where vendors have recommended specific hardening practices or options, those that are being applied should be referenced in the plan. Where no vendor recommendations are available, an IT evaluator should examine the system or device (this can be done during a Proof of Concept test), to determine which network practices and protective measures should be applied and included in the plan.
© 2010 Bp.IP Initiative