IT organizations everywhere are under pressure to reduce TCO (total cost of ownership), however TCO is no longer dominated by hardware acquisition costs. Yankee Group reports that ten years ago businesses spent two thirds of their IT budgets on product acquisition and one third on managing their IT environment. Whereas in 2006 they spent 70% to 80% maintaining existing environments.

And these management and administration costs are ramping, as is shown clearly from IDC's projections on New Server Spending and IT Management Costs. IDC identified this ramp was driven by inefficient use of management resources, downtime and associated costs. The solution to this growing problem lies in improved IT infrastructure management, and one key to significantly reducing infrastructure downtime and operating cost is Out of Band (OoB) console management.



Console servers are one of the key system administration tools required to keep today's complex, distributed networks operating with maximum uptime. When properly configured they enable disaster prevention, maintenance and disaster recovery for a wide range of computing equipment, communications systems and other critical infrastructure. Console servers connect to a network in-band, usually via trusted Ethernet network and also connect to the administration console of each system being monitored, controlled or maintained. In order for console servers to allow access even when the network is inaccessible, they usually also are configured with an alternate out-of-band (OoB) network connection (via a dial-up modem or ISDN). This is vitally important because if there is a central network outage, then the console server is still independently accessible, allowing the system administrator to maintain remote systems and get them ready for recovery. These systems include- server computers (tower, rack optimized, blades), network infrastructure (routers, switches, and firewalls), communications systems (PBX, etc.), environmental (air conditioning, heating, lighting) access and security systems and power distribution systems.

A system administrator can potentially access any system's console attached to serial ports on the console server, from their management station. This is achieved using a secure network terminal program such as 'SSH' (secure shell). It is a secure alternative to 'telnet' and unlike 'telnet' provides strong user authentication (certificates) and data stream encryption (AES, 3DES, etc). Most console servers also have access control lists and administrator authentication based on a wide range of industry standard schemes such as RADIUS, TACACS+, LDAP and so on. In addition, administrators may have different permission profiles so that they can only access and maintain certain equipment or certain servers attached to particular ports.



For simplicity and flexibility, console servers use a standard RS-232 serial port to connect to a console or service port on each individual system (router, firewall, server, power strip, PBX and so on). They are a true multiplatform management appliance. Console servers are sometimes packaged in desktop enclosures (8 ports or less) for small sites and rack optimized enclosures (16 to 48 ports) for dense sites such as datacenters and co-location facilities, which may house large server farms.

A serial console port usually transmits terminal-like text screens. Even Windows 2003 and XP server now offer an administration console on the serial COM port (with EMS); and the BIOS boot sequence likewise can be redirected to a COM port. Serial ports provide a number of advantages over graphical displays when it comes to IT infrastructure management. One advantage is that serial ports have a low data rate (9600 bps to 115200 bps) so even if scores of console ports are streaming messages they will only consume a fraction of the Ethernet bandwidth. Therefore console servers can operate adequately even over narrow band connections such as analog and ISDN modems. As a result they can be placed almost anyway on the network with minimal bandwidth impact.

An alternative remote management option is to use graphical solutions based on remote control programs (like PC Anywhere and VNC). These are excellent local operations tools however for system administration they require the system being managed to be stable enough to run the program. If a major subsystem is unreliable or unresponsive, such as the network stack, then access is lost and the administrator may need to physically access the remote site. These tools also can easily monopolize network bandwidth as bit mapped screens contain megabytes of data. Another common management tool is KVM over IP. While this also consumes volumes of network bandwidth (even if compressed they still generate orders of magnitude more data than serial consoles) they do provide system administration access even when the system being managed becomes unreliable or unresponsive.

A newer management solution is the RDP (Remote Desktop) manager which is now an integrated part of all Windows systems (and being supported by SUN and others). RDP provides the remotely connected user (or administrator) with a graphical interface connection, while minimising the data flow over the link to that user. Remote users can have a direct RDP connection to the managed server over dial up modem links. Also some console servers now support remote secure RDP connection through COM ports.

Also there as many other infrastructure devices that do not contain a graphical management capability, or a network connection, but that do have a serial console port. These ports transmit machine-readable data streams. That means the console server can capture the screen messages across the serial port and log them into internal storage or onto external storage on another server on the network. This is not possible with bit-mapped graphical displays as they are only for visual display information.

Console message history can be used for disaster prevention, maintenance, forensics and recovery. Most console servers can scan data streams looking for patterns and send emergency alerts (e-mail, pager, etc.) to system administrators or technicians based on specific errors. Some examples include, running out of disk space, tape backup failure, application failure, and fan failure on a blade chassis and so on. Even if it was not possible to avert a disaster, often forensics on logged information can lead to better prevention in the future. In addition serial console servers can transmit specific messages to mission critical devices such as power distribution strips allowing remote power cycling of individual servers as well as power and temperature monitoring.

So although it is not the only tool available for remote system administration, serial console servers are the best tool. They reduce the TCO of infrastructure management because they can manage the widest range of mission-critical business equipment. Their ability to effectively provide for disaster prevention, maintenance and disaster recovery, with a low impact on network resources in one simple appliance makes them an imperative towards ensuring the maximum business uptime.