Console Server Technology - Industry Trend or Event

Computer Technology Review, Dec, 2000 by Mark Prowten

Solving the critical problems of remote network management

With organizations relying more and more on the network infrastructure as the primary conduit for all information, the demand upon network managers to provide a network that is free from downtime and performance constraints is greater than ever. Couple this demand with the fact that many organizations are using wide area networks which span many remote sites or campuses, have multitudes of remote users dialing in from various locations worldwide, and are supporting connections to the Internet for advertising, e-commerce, or for data gathering. Increasing sales in network infrastructure equipment indicates this trend towards expansion. In fact, the Dell 'Oro Group, analysts of the switch and hub marketplace, has reported that sales of network switches will grow to an excess of $23 billion by year 2004 (an annual growth rate of 20% per year over the next five years).

Organizations with multiple sites will have a myriad of switches, routers, hubs, access servers, and networked computers such as PCs, workstations, file servers, and even some mainframes. The key fact here is that these devices are in a number of locations and, depending upon the size of the organization, could be located across the country or even continental boundaries.

Today, network managers are on the hot seat to maintain both availability and performance of the organization's network regardless of how widely dispersed the network infrastructure might be. Generally hindered by small staffs or by a lack of seasoned personnel, time is a network manager's most valuable resource.

The stakes are high: Infonetics Research recently reported in a study of Fortune 100 and Fortune 500 companies that network downtime totaling only 5.5 hours for the average company could cause a loss of up to $3.9 million in revenue and productivity. The problem is how to expand the capabilities of network management personnel within an organization to better maintain the variety of network infrastructure devices and networked systems presently being used to minimize and possibly avoid network downtime or performance loss.

In-band Management--An Incomplete Solution

At first glance, when analyzing the problem of how to better manage a widely-dispersed network infrastructure and networked systems, a network manager might decide to purchase only those network switches or routers that support in-band management via SNMP, telnet connections, or proprietary management software. The idea of having the entire network mapped onto one or more monitoring servers sounds appealing and potentially simple. However, this strategy has a number of flaws:

In-band management as a solution is limited. It relies on the existence of the very network which one is intent upon managing. So if your only means of management is through network connections and then the network goes down, you are left without management capabilities.

SNMP (Simple Network Management Protocol)

SNMP provides limited abilities to manage devices using a standard developed to run on the IP protocol. The advantage of SNMP is that it can be used to manage devices supporting the protocol from any number of hosts or PCs running SNMP software. The limitation is the same as any other in-band management tool: what to do if the network itself has failed?

Telnet

Telnet is of course one of the great features of the IP protocol stack in that it allows connections just about anywhere but sometimes users are limited to a subset of features by certain devices. Many vendors want to protect users from themselves by not allowing them to change the IP address or reboot a device that they are connected to.

Vendor-specific software

While some vendor-specific packages provide a wealth of information beyond that of ordinary telnet or even SNMP, the cost/benefit decision remains. The specific features a software package provides may not be fully realized in the hands of a relatively untrained staff.

Devices without network ports

When one considers all of the devices which might show up in a computer rack, there are a number of devices which might be critical and yet have no network port at all (PBXs, UPSs, cooling systems). For these devices, the only access is via a console port connection.

So a simple strategy based on in-band management will provide some assistance in certain circumstances but will have minimum benefit if a key router goes down or if there are a number of critical devices which do not have network ports. A simplistic solution to this problem is to connect those devices which are "really critical" to modems for out-of-band connectivity. But will this strategy make sense in rapidly expanding networks? If there are 50 "really critical" devices, then this could entail the installation of 50 modems and 50 phone lines-a very expensive and potentially difficult-to-maintain configuration.

Console Access--The "Proximity" Issue

In the 1970s and 1980s, when many organizational networks were LANs rather than WANs, the central computer room was the focal point of all information processing. Various mainframe or mini computers supplied most of the computing power and much of the storage devices were either large disk systems or magnetic tape drives.