Larry and ALL, interesting article... Intel's LAN servers...
January 15, 1997
Helping Pentium Pro servers keep their cool
By Mark Stanczak, Pc Week Labs
ÿWhat would it take to turn a network administrator into a network prophet, someone keenly aware of each component within any given server and able to respond before a minor problem escalated into a critical loss of service? The Pentium Pro-based servers destined for corporate LANs in 1997 could hold the answer.
Many of the latest midrange and high-end Pentium Pro systems incorporate specialized components such as thermal sensors and CPU-utilization monitors for tracking the health of the server. These features act together to create an early warning system that detects component-level failure and alerts network administrators before severe damage can bring a server to its knees. With such protection in place, today's Pentium Pro-based file servers clearly offer more than just an improvement in processing power.
The greatest enemy of any file server is high temperature. At 3.3 volts, the Pentium Pro draws a good deal less power than early 5-volt Pentium processors, which suffered from severe overheating problems. Nevertheless, today's Pentium Pro processor generates a significant amount of heat during normal operation.
To make matters worse, the latest 7,200-rpm disk drives included with most Pentium Pro servers run hotter than ever before; the upper operational limit for these drives is 122 degrees Fahrenheit. Combine four Pentium Pros with a stack of fast hard drives plus a pair of power supplies, and the resulting system could perform double duty as a small room heater.
Only myriad fans and ducting designed to move cooler outside air in and hot air out protect the server components from their harsh climate. Failure of this environmental control subsystem will allow components to overheat, eventually causing a server crash and requiring replacement of the fried parts.
To address these issues, the Pentium Pro motherboards manufactured by Intel Corp., Compaq Computer Corp., Dell Computer Corp. and Advanced Logic Research Inc. use advanced status-monitoring technologies to notify network administrators of major component-level failures.
For example, should a fan that circulates air around the disk drive bay slow down, the disk drive hardware detects the speed reduction, small thermal sensors perceive the subsequent temperature rise and the facilities built into the motherboard report the alert condition. This notification can be sent to a management console such as Novell Inc.'s ManageWise over SNMP, directly to a service center using a locally attached modem and dial-up link, or to a technician's pager.
Taking proactive warning systems one step further, Data General Corp.'s Aviion and Stratus Computer Inc.'s Reliable Architecture Distributed I/O Cluster nearly fix themselves. These systems immediately notify the server manufacturer--via the LAN or dial-up link--of a component failure so replacement parts can arrive on-site within 24 hours via express delivery.
______________________________________________________
I see this as a very important market for the next decade.
Regards, Michael |
