When designing a new data centre, it is good practice to look closely at
various critical components to mitigate the exposure to overloads, heat and
leading power factor, writes Christelle Larkins, area manager: South Africa,
MGE Office Protection Systems. In addition, users must be aware that power
management becomes more critical than ever in a high density environment.
Non-blade computers require components that are bulky, hot, space
inefficient and duplicated across many computers that may or may not be
performing at capacity. By locating these common services in one place and
sharing them between the blade computers, overall utilisation is more
efficient. The specifics of which functions are included and how those
functions are implemented vary by blade vendor, but they share common
characteristics that impact the design of today's data centres, especially
in terms of power and cooling.
Although blade servers have many benefits for some applications in terms of
space utilisation and potentially improved reliability, the very nature of
high-density computing results in issues that must be addressed throughout
planning, implementation and operation. Proper consideration of added
cooling and power demands as well as changes in power delivery will assure
that users achieve the optimum levels of availability with this new
As part of the strategy to improve overall data centre reliability, most
servers have moved to dual-corded power input design to increase the "nines"
of availability by eliminating single points of failure in the event one
power source or cord failed.
However, this has the obvious but often ignored consequence of lowering the
load on each power supply. In transmitting only 50% of the power,
efficiencies decrease, cooling and energy requirements increase and problems
arise with leading load power factor.
Most traditional UPSs were designed for 0.7 to 0.8 power factor loads and
must be de-rated by as much as 30% to support a blade server load of 0.9
leading power factor. Fortunately, major UPS manufacturers have recognised
these changes and responded by introducing blade server-friendly UPS systems
that do not require de-rating.
Typical issues created by blade servers include:
* Higher power factor loads that result in greater Watts per VA;
* High inrush current;
* High heat density (3kW to 5kW) from one blade centre;
* Higher power use in a given area due to greater density of demand; and
* Dynamic server demands that grow and change based on use and applications.
These blade architecture considerations call for use of flexible power
protection hardware that can offer the most responsiveness in a dynamic
power demand environment. Traditional UPS systems must be oversized to
address these issues. However, MGE Office Protection Systems' Pulsar MX UPS
models have been designed specifically with blade architecture in mind. They
can be installed in rack or tower configurations and offer modular designs
that can accommodate changing server configurations, redundancy and backup
time requirements that can provide up to 20KVA or 15KVA in redundant
configurations using 5KVA sub-modules, while hot-swappable components enable
maximum uptime, even during maintenance operations.
When designing a new data centre, it is good practice for those responsible
for planning and implementation to consider these factors and take adequate
steps during selection and sizing of various critical components to mitigate
the exposure to overloads, heat and leading power factor.
It is vital that the IT team, facility management and equipment suppliers
work together to create a total solution that will maximise reliability,
availability and system efficiency while optimising installation and
operation costs. Fortunately, solutions are available to simplify this
process and enable use of blade architecture to its fullest advantage."