Server clusters can avoid the worst effects of crashes.
Unlike familiar desktop and notebook PCs, which support one user at a time, server computers play host to hundreds of people. When a server experiences a crash or data corruption, the problem affects an entire community. The severity of the outage depends on the type of problem encountered and the ability of the server's hardware and software to handle the problem.
Physical Effects
Physically, most server failures are subtle: you might never know something was wrong by the computer's appearance. A few effects, however, are evident and dramatic. Worn hard drive or cooling-fan bearings make a metallic grinding noise. A server motherboard with leaky capacitors has drips of mildly corrosive electrolyte or will may emit puffs of black smoke. Heat or voltage-spike-damaged motherboard components may char or explode. A server's microprocessor overheats and burns if its cooling fan fails.
Users and Processes
Once the server's microprocessor stops working, users logged into the machine lose access immediately. If the failure corrupts data on hard drives, the users may suffer the loss of an entire day's work. A server providing Web hosting, for example, can no longer send Web pages to users. It will also lose any independent or batch processes it had in progress, such as backup operations, large data transfers or automated database jobs.
Data
The entire contents of a hard drive may become lost due to corrupted data. A hard drive's data directory and file structures are normally very orderly; depending on the degree and location of data corruption, the drive may lose only unused free space or it may lose a handful of files. If the corruption occurs in the hard drive's boot sector or file directory, it could make the whole drive unreadable. The computer's main memory is volatile: if the power goes off, the memory loses its contents. A hardware problem involving the loss of power wipes out the data in main memory.
Loss Prevention
An organization that practices daily file backups loses at most those files that changed since the last backup. After the server's repair or replacement, the computer staff copies files from backup media back to the main hard drives, restoring the computer to nearly the state it was in when the problem occurred. Organizations that spend money on extra technologies may avoid data loss and downtime. A technology called RAID-5, for example, stores data redundantly on several hard drives. Technicians recover from the failure of one drive in a set by simply replacing the drive; RAID-5 automatically rebuilds all the data in the set. In a similar way, server-clustering technology avoids downtime and data loss through redundancy: the failure of any one server computer in a cluster does not bring the system down.
Tags: hard drive, hard drives, data corruption, data hard, data loss, main memory, server microprocessor