|
Sometimes, we're justified in saying ignorance is
bliss. But only when we're in a position to assign the
burden of knowledge to someone else. In the
communications solutions space, many of us have
achieved a sort of bliss by letting someone else worry
about the quality and availability of electrical
power. Such responsibility was usually delegated to
the telecom guy, a cooperative sort who tended to the
hulking crates in the telecom closet, or
responsibility was diffused somewhere within the PSTN
cloud, a vague, misty realm where esoteric notions,
such as "five nines" reliability, drifted by at an
agreeable hazy distance.
Unfortunately, the pleasures of displacing power
anxieties are no longer so easily available, not when
convergence proceeds apace, when enterprise telephony
becomes just another mainstream IT function, and the
public network seems less and less monolithic, but
rather an agglomeration of circuit-switched and
packet-based hybrids.
Overall, these changes suggest a trend towards
independence, manifest as freedom from proprietary,
mainframe-style PBXs or as a proliferation of
specialized service providers, and an ever richer
assortment of enhanced services options. With
independence, however, comes responsibility.
VOICE, DATA, AND THE FORGOTTEN NETWORK
With so much speculation about the convergence of
voice and data networks, little attention remains for
contemplating the role of the power network. And yet,
as communications systems evolve, so do their
corresponding power requirements. Let's consider how
power requirements may relate to each of the following
communications trends:
- Instead of a solitary key system or PBX, a
corporation may deploy and manage a constellation
of voice-enabled servers, routers, and gateways.
And individual desktop computers may implement
each user's personal configuration and access
preferences. In short, the trend is towards
accommodating a wider array of network elements,
and incorporating them in more distributed, more
complex network architectures, which may leave
corporations more vulnerable to adverse power
events.
- Corporations are increasingly likely to
consolidate voice and data functionality by
deploying premises-based integrated access devices
(IADs), PC-PBXs, and IP-PBXs. While convenient,
such devices could bring down both voice and data
if they were to fail, thus increasing the stakes
of any power failure event. Another stake-raising
communications application is unified messaging,
which may involve a single server consolidating
the functions of voice mail, e-mail, and fax
servers.
- More corporations and contact centers are
embracing multi-channel communications, not just
in the sense of deploying alternative independent
channels, but in the sense of orchestrating a
variety of channels, accommodating each customer's
contact preferences, translating freely from
medium to medium, depending on content, time of
day, nature of the communication (notification,
reminder, etc.), as well as the perceived value of
the customer relationship. Such intricate plans
could come to naught, however, if even a subset of
channels (or even a single channel) were
unavailable due to an adverse power event.
- Increasingly, broadband data access is
accommodating real-time, mission-critical
communications. Telephony-ready cable
infrastructure, DSLAMs, wireless data
infrastructure, and gateways of various types
increase the number of end office elements with an
exceedingly low tolerance for power-related
service interruptions.
- Corporations are weighing the advantages and
disadvantages of outsourcing, often to data
network providers of various types, many of which
plan to carry real-time, mission-critical traffic
on behalf of subscribers. Before committing to any
outsourcing scheme, a corporation will want to ask
about the provisions for power reliability the
service provider may have implemented. And, of
course, the service providers will need to have a
compelling power control story.
For providers, something as basic as access to
reliable power could become a competitive
differentiator.
FEARFUL SYMMETRY
It is curious that just as voice/data convergence
increases the general reliance on the power grid, the
limitations of the grid are growing more obvious. The
grid, while fairly well adapted to the needs of
manufacturing and residential consumers, is less well
suited to a high-tech service economy that relies on
sensitive electronics. While a drill press or a hair
dryer may function well enough despite occasional
power fluctuations, circuit boards controlling
communications are fussier. They require precision
power. Plus there are issues of availability.
Availability, as the extreme example of California
attests, may become less rather than more assured,
particularly if provisions for growth lag behind
actual demand, or if policy missteps derange the
mechanisms of distribution.
If traditional utilities prove less than completely
reliable, or deliver power that is less than constant,
alternatives such as self-generation or co-generation
become more attractive. However, these alternatives
may deliver power that is at least as "dirty" as that
from a utility, and switching from one power source to
another can itself introduce disruptions. However, a
general trend towards more local power might also
favor a trend towards greater reliance on DC power,
since it was the need for efficient sharing of power
over long distances that justified reliance on less "clean"
AC power.
CONVERGING POWER PRIORITIES
In the access network, in the end office, and
elsewhere in the telco space, the emphasis has always
been on ensuring maximum uptime, providing nearly
uninterrupted support for real-time, mission-critical
communications. It was, after all, in the telco space
where commitments to six or seven "nines" of
reliability evolved.
Similarly evolved commitments apply in the IT
space, at least in large data centers, which may even,
like the telcos, rely on DC power, at least for
certain classes of equipment. However, in many IT
scenarios, power solutions tended to emphasize
equipment protection, perhaps via a simple, disposable
surge suppressor, or perhaps via relatively
short-lived battery backup, which would at least
permit a managed shutdown, if not continuous operation
through an adverse power event.
While controlled shutdown capabilities are and will
remain important, they do prompt the question of how
much downtime is acceptable, especially since more and
more IT functions are deemed mission-critical, and
since IT may find itself responsible for maintaining
converged voice/data infrastructure, which supports
communications flows which are inherently intolerant
of interruption.
Ultimately, we may see a convergence of power
priorities. While those maintaining communications
infrastructure may enhance uptime capabilities, in
emulation of telco practices, they may also emphasize
management flexibility, exercising discrimination
about what must stay up, and what may be shut down,
with an eye towards extending battery power though an
adverse power event. That is, we may see more emphasis
on self-diagnostics, manageability, and serviceability
in all kinds of communications equipment, including
classes of equipment for which such discrimination had
scarcely been thought necessary.
Another area of converging priorities involves
power density, which is closely related to yet another
common concern, namely, cooling. Power density, or the
amount of power consumed within a given volume, grows
as computing and processing power grows, and as
equipment becomes more compact, despite delivering
more performance. Power density increases indirectly
as a result of Moore's Law, and directly as a result
of the increasing value of central office, data
center, and collocation space.
The more power that is consumed within a given
space, the more heat is generated, and, accordingly,
the greater the need to dissipate heat. In the absence
of effective temperature control, excess heat can
shorten equipment life or even instigate a system
failure.
POWER SOLUTIONS
In keeping with the general idea of converging power
priorities, telco-oriented operations will be mindful
of how much more sensitive digital circuitry is
compared to analogue circuitry, as well as more
demanding in terms of scalability, manageability, and
serviceability, particularly in the light of greater
competition amongst service providers and the
proliferation of network equipment, including the
possibility of service providers assuming greater
responsibility for customer premises equipment.
IT-oriented operations will weigh their power
control options while remaining mindful of how
business imperatives place less emphasis on mere
equipment protection, or managed shutdowns, and more
emphasis on continuous uptime.
Both telco- and IT-oriented operations may be more
inclined to rely on hybrid AC/DC systems. Telco
operations will continue to rely predominantly on DC
power and 48V systems, and IT operations will continue
to evolve their AC UPS approaches to power control.
Accordingly, on either side, the advantages of
maintaining a single infrastructure, AC or DC, will
outweigh the disadvantages of additional power
conversion. For example, an IT operation responsible
for some DC-reliant communications equipment would
probably opt to convert some AC power to DC, instead
of installing, managing, and maintaining both AC and
DC batteries.
As for UPS systems in general, we could hardly
conclude without first quickly reviewing the three
basic UPS types: standby (or offline), line
interactive, and online. They differ, for the most
part, in how seamlessly they accomplish the switch
from generated power (from the utility or
self-generated) to stored battery power. The least
expensive option, standby, accomplishes the switch
more or less precipitously, upon detection of an
adverse power event, a brownout or blackout. The
somewhat more sophisticated line-interactive option,
accomplishes the switch more delicately or
selectively. That is, a line-interactive UPS will rely
on a voltage regulator to adjust power up or down to
accommodate brief sags or surges, resorting to backup
power only when voltage exceeds a predefined range.
Finally, the most expensive UPS option, online, doesn't
switch at all. All current is converted to battery
power and conditioned, eliminating sags and surges and
noise, before powering equipment.
CONCLUSION
Ignorance of power requirement is no longer tenable,
for such ignorance no longer a source of bliss, but
inevitably, pain and loss. Instead, bliss should
reside in knowledge, that is, in knowing you've made
appropriate provisions. Certainly, this is the sort of
bliss appropriate to power control.
|
|
While shedding complacency and
building more sophisticated communications
networks, IT- and telco-oriented operations will
understand that power control issues have been
influenced by the overall trend of convergence.
Both will recognize that reliable power is about
more than equipment protection, but is, rather,
about the integrity of uninterrupted data flows
and preventing productivity loss to downtime.
Both will be sensitive to the need for finely
scaled deployments and sophisticated management
capabilities, achieving sufficient but not
extravagantly excessive power control capacity. |
[
Return
To The August 2001 Table Of Contents ]
|