Serviceability

Let's say that as a network administrator you find out that nothing is being received over a specific network node. Investigating, you find an I/O card has gone out on the system. On a typical Itanium-based system, you can pull out the card while the system is still running and plug another one in—it's 'instantly serviceable.' You don't have to shut the machine off.

Hewlett-Packard computers typically have system component redundancies for subsystems like fans, and power supplies. We tend to build the redundancy into the areas that statistically tend to have the most failures. Thus if the fan fails, the entire unit doesn't burn up. Similarly, built into the architecture is the enabling capability to allow for OLR (Online Replacement). This is another way of saying that you can pull out a component (e.g., I/O card) while a system is running, and keep on going. Instead of engaging in costly downtime, replacement can be done on the fly.

This is a capability that has been built into the architecture. In a multi-processor situation, you can turn off one processor and substitute in a new one. The ability to service the machine while it is still running reduces your down time even further. It's rather like being able to replace a flat on your car while in the middle of your freeway commute.