|
 |
It is almost impossible to ensure
that a UPS component never fails. With the PowerReactor3's parallel redundant
architecture, however, any single component failure can be prevented from
affecting the functioning of the whole system.
All modules in the UPS run in parallel.
The load is evenly shared by each of the modules. Should any module fail
or be removed, the remaining modules continue to function and take over
the load. Load-sharing occurs automatically and instantaneously.
The PowerReactor3's proprietary
Multi-Master technology allows the modules to function without a central-command
unit. This greatly increases the redundancy as the redundancy is multiplied
by the number of the modules remaining in the system. The
more modules in a system, the greater
the redundancy. No need to rely on just one or two central-command units
for fault-tolerance. |
Fault-Tolerance Needs More
Than Two Brains
Some claim that having two UPSes
in parallel will provide adequate fault-tolerance, as one will back up
the other if one fails. Others argue that having multiple power sections
controlled by two central-intelligence modules will provide adequate fault-tolerance.
However, such scenarios are equivalent
to decision-making by a two-person committee. When one disagrees with the
other and each claims the other is a failure, a stalemate occurs, and no
action is taken.
To properly achieve fault-tolerance
in a UPS, an effective voting system has to be established. Such a system
must be designed to depend on more than two decision-makers. In a PowerReacfor
UPS, the decisions of all the modules go through a voting process and the
final decision lies with the majority. |
Fail-Safe construction of a parallel redundant
power system
In a PowerReactor UPS, each module
has its own set of components and functions independently of the other
modules. Each set of components needs to bear only a fraction of the rating
of the whole system. If one component fails, it will only affect its own
module. The other modules in the system will continue to function.
Example of a typical set-up used in conventional
UPS
In a conventional UPS, each component
is responsible for a specific function. Each component needs to be sized
for the fully rated output of the UPS. When one fails, it can cause the
whole UPS system to fail.
|