| Remote physical security and reducing downtime in the data centre
INTRODUCTION Generally speaking, the IT infrastructure is
exposed to two main types of risk:
The role of any security strategy should be to evaluate the above risks and reduce their potential impact on the company’s IT assets as much as possible. For example, threats of outside attacks should not be ignored, but the above figures indicate that human error may have a greater impact on IT performance. To reduce these threats, IT administrators in the data centre environment should implement a physical security plan that makes servers accessible to only authorised personnel. This rigid approach limits the type of access available to individual users and provides administrators with a greater level of control. SECURITY STRATEGY COMPONENTS Physical security These work environments are uncomfortable and insecure. They also result in higher support costs from reduced productivity; and increased risk of personal injury from rolling a cart through and around racks. The components of physical security are
as follows: IT management is increasingly facing growth in very controlled environments, especially as data centres increase in size and are in different geographic “lights out” locations. Authorised personnel may be required to enter and exit using special issue access cards and biometric finger scans. Video surveillance cameras in the building may monitor all activities in strategic locations. The question for many IT managers is how to supplement physical security strategy. The answer is to give secure, remote access and control of data centre servers and devices to authorised personnel no matter where they or the devices are located. Logical security Logical security includes defining and protecting resources. What resources can users access when they have been authenticated? Logical security also involves defining the administrative authority. Who has the authority to administer both user passwords and their levels of access? An important issue in managing servers and devices is that some may have their own unique management interface, authentication and password lists. There is no centralised user access management between systems. It would be of benefit to the administrator if logical security for all devices could be managed from a single platform. Auditing and reporting There are a number of suspicious activities that constitute security events and administrators will know to be wary of the following: * System access denied Another useful facility in administrator audit mode is stealth control which enables the administrator to watch activities and changes on a server or device in real-time, without the user being aware and take immediate action to disable the user, if required. Presently, many systems will not allow the administrator to cross reference audit information between different devices, requiring them to manually collate the information into a usable report. In these circumstances, a common interface for logging all user access and a widespread audit trail would be of great benefit to the administrator. Alert management The alert system should support SNMP MIB-II for integration with an existing enterprise management tool such as Tivoli, OpenView or Unicenter. The SNMP traps should be based on user-defined levels. User applications influence physical
security Test labs where multiple users simultaneously access and monitor testing simulation on hundreds of servers. Access to such servers is critical because testing simulation typically requires a large amount of hardware, software and integration testing. Demo labs where multiple users need to monitor
live testing on hundreds of servers from various locations throughout
the lab. Apart from physically securing the servers, the real-time access scenario also has requirements for logical security, auditing and reporting, and alert management. The second user access type is administrative-level access. IT administrators are faced with the daily challenge of managing many different and distributed systems across the enterprise. In many cases the administrator needs to have full administrative-level access to servers and devices no matter where they are located. Every administrative function conducted on a target device, including full-power recycling, watching screens as a machine boots-up and access to BIOS settings, needs to be conducted as if the target device is in the same room. This level of access is required both on locally and remotely. It is important that the administrator controls all administrative-level access to servers and devices. The management platform should secure all device access by leveraging the directory system already in place, allowing the administrator to use a single user name and password repository. Device-level rights must be assigned based on a user’s name so that administrators have access to more devices than, for example, an entry-level technician. Auditing and reporting of all activity is also important to keep track of “who does what” within the network infrastructure. The challenge is more complex when administrative-level access to secure servers and devices is required from multiple remote locations. Not only should the links be encrypted and secure, the actual access needs to be controlled through common authentication and tracking procedures. THE ROLE OF KVM IN ENHANCING PHYSICAL
SECURITY What is KVM? This is facilitated by connecting directly to the keyboard, video and mouse (KVM) ports of target devices, enabling operators and users to access multiple computers as if they were sitting directly in front of each machine. Although many server management functions can be performed remotely through network management systems tools such as Tivoli, OpenView or Unicenter, some more basic levels of server configurations can be accomplished only through KVM access. An example of this is the boot process in which the BIOS and the operating system configuration and set-up take place. This process occurs before the networking layer is operational on the given server, so network-based tools cannot be used. In recent years the scope of KVM technology has expanded beyond the local control of multiple racks of servers over a proprietary network. KVM control of target devices from any location is now available over standard protocols such as an IP network. KVM technology has expanded to allow access and control of serial devices such as headless servers, routers, power strips and environmental systems. Additionally, administrators can now maintain and troubleshoot all their servers and serial devices from anywhere using one screen and management software. The simplification of management through a single-seat scenario allows for better management of firewalls, host and network-based intrusion monitors and access control – vital parts of the security landscape for any data centre manager. A single-seat KVM over IP solution provides secure access to servers and network devices from a single software interface, allowing the same common software to be used to access and control other layers of security such as firewalls and network-based intrusion monitors. A KVM solution enables a proactive – as opposed to reactive – security approach to data centre management. This is the logical step forward for most network administrators. It helps afford significant long-term cost savings in comparison to some of the more reactive solutions such as pure intrusion detection – where potential hackers may already have access to the network. About Avocent
•Date:
3rd August 2004 •Region: World •Type:
Article •Topic: IT
continuity |
