Data centre monitoring

and management

e book
www.ECDsolutions.com.au
 W hether its the server under the stairs or the
worlds largest - the 2,200,000 square foot
SwitchNAP facility in the Nevada desert -
the data centre is the hub of any business. In
many cases, it is the business itself.
Increasingly, data centre security, power consumption and
environmental monitoring are considerations on the radar for
administrators. Data centres are susceptible to hits from all angles:
viruses, spyware, network threats and cyberattack, heat, humidity,
airflow and smoke. In fact, its pretty frightening how quickly and
easily a companys IT operations can be taken down, which makes
it hard to understand why some companies do not implement a
solid strategy for dealing with IT threats.
A combination of environmental and security monitoring,
remote console management and remote power control gives
administrators the power to control data centre environmental
conditions and security, server and network equipment,
and power distribution and usage. Automatic notification of
environmental conditions or security breaches before the event
disrupts network infrastructure can mean the difference between
a minor inconvenience and a major crisis.
Couple rising power prices with the need to closely control the
environment and data centres can easily be seen as the bad guy
when it comes to the overall energy consumption of a business,
but this neednt be the case. A bit of forethought in terms of design,
a few behavioural changes and a system of monitoring and
controlling power distribution can help keep energy use in check.
Todays data centre has an expected lifespan of just under 20 years,
so making a few informed decisions and considered investment
now can make all the difference a decade down the track.

Dannielle Furness
Editor ECD Solutions

Contents
3 Remote console management and power switching

5 Energy use and data centres

7 How to remotely reboot after a system lock-up

8 The data centre of the next decade

11 Increasing uptime with improved environmental monitoring

 iStockphoto.com/Anthony Hall
Remote console management
and power switching
Stephen Withers

Once upon a time, almost all of an organisations computing equipment would be

in the same place as those responsible for keeping it running smoothly. Todays
practices include locating significant pieces of equipment in branch offices where
there may be no IT staff and taking space in third-party data centres (which
makes physical access less convenient) instead of operating a private facility.

F
ortunately, a number of technologies are from a centralised console, though not all such
available to overcome or at least minimise systems are able to manage all types of devices.
the negative effects of physical separa-
This approach is known as in-band management,
tion of systems administrators from the
because the management traffic flows across the
hardware they manage. Among them are remote
same network as the rest of the data.
console management and remote power switching.
This is all very well as long as the network is
running properly, but what happens when a device
Remote console management
such as a switch or firewall is at least part of the
A basic part of the picture is remote console problem and therefore network communications are
management. interrupted? Thats where out-of-band management
comes into play.
Many network or networked devices - servers,
printers, routers, switches and others - include a Many devices are still fitted with a conventional
web-based console allowing remote management serial port to allow the connection of a terminal
and configuration. In addition, or alternatively, (these days its more likely to be a notebook running
they may make provision for remote management terminal emulation software rather than a dedicated
via other network protocols such as SNMP, Telnet terminal) for configuration or management. While it
or SSH. is possible to attach a modem to the serial port to
allow remote management, that is rarely practical
Apart from manual control using such protocols,
where multiple devices are concerned.
there is a range of systems management tools from
3
various vendors that allows device management Enter the serial console server.
 A serial console server connects to the serial latter being a particularly useful way of maintain-
ports of multiple devices and is in turn connected ing service levels, though the underlying cause of
to a modem (for dial-in access) or via a completely recurring failures should be investigated). These
separate IP network. This allows staff at other loca- and other conditions such as power supply ir-
tions to dial in (literally or metaphorically) to the regularities may also be reported via email or to
console server and connect to the troublesome a syslog or SNMP server.
device. This means it is important that the console
More advanced models may include power
server has appropriate security features such as
metering at the aggregate or individual circuit
strong authentication (eg, via a RADIUS or Kerberos
level, as well as metering the actual input voltage.
server) and logging - and, where dial-in modems
are used, dial-back capability to ensure the call Remote power switching can also be used to turn
is coming from an authorised location. devices on or off when they are needed (possibly
according to a schedule) to save power. In more
If the console server is not accessible via the
complex environments they can also be used to
network (which is the main reason it would be ac-
balance power loads across the machine room.
cessed via the serial port), then the usual network
security measures would not be available and Remote switching is not only for devices that
so local authentication would be required. Other run on mains power. Models are also available
security measures include keystroke logging and for remotely switching DC power, including 12, 24
automatic screen captures - such features do not and 48 V supplies. Such DC systems are said to
reduce the risk of intrusion, but they do add to reduce energy losses and therefore achieve greater
the audit trail. efficiency, as well as being simpler and cheaper
than AC power distribution systems.
Another issue is that the console server should
be able to check that an attached modem is Not all remote power switches are rack mounted:
functioning correctly. If a simple re-initialisation some models resemble an oversized power strip
fails to restore normal operation, rebooting via a and are designed for vertical mounting within a
remote power switch (see below) should do the cabinet and others are packaged in standalone
trick. The server should also periodically check for cases to support a small number of devices that
a dial tone. Given that the modem is the alterna- are not rack mounted (eg, a small branch office
tive way into the console server in the event of server or a kiosk in a shopping centre).
a network failure, it is important to attend to any
Both mains and DC remote switches may also
communications issues promptly and as far as
provide automatic switching to a secondary power
possible automatically.
supply should the primary supply fail or become
Console servers are not limited to out-of-band unstable. Combining automatic switching with
communications via serial interfaces. An alternative remote switching in this way saves rack space.
arrangement is to configure a completely separate
Another way to save space - especially in smaller
IP network for communication between the console
installations such as branch offices - is to select
server and the central control point. Such a fall-
a device that combines the remote console server
back network is much simpler than providing a
and remote power switching in one box.
secondary network path for every device, but if
any components are shared with the main network As when choosing other pieces of network
it may not be possible to reach the console server equipment, it is worth checking that a remote
in the event of an outage. power switch supports IPv6 as well as IPv4, as the
former is likely to come into common use within
the lifespan of the device.
Remote reboot power switch
Sometimes a device is completely unresponsive,
even on its serial port. In that situation all that is left Conclusion
is to cycle the power. How do you do that without Remote console servers and remote power switches
gaining physical access to the device? Use a re- make it easier for IT operations staff to deliver
mote reboot power switch, also known as a network improved service levels, especially where the equip-
power switch. Such a device can be thought of as ment they manage is geographically dispersed or
a remote-controlled power strip. Typically mounted located at a third-party data centre with restricted
in a rack along with the equipment it controls, the access. Remote console servers make it easier
switch provides multiple independently controlled to manage multiple devices with serial console
mains sockets that can be turned on or off via its interfaces and may provide a backup control path
network or serial interfaces. For maximum robust- in the event of a network outage. Remote power
ness, that serial interface can be connected to a switches allow power cycling without the need for
serial console server along with the other devices. physical access to the hardware.

Remote switches may provide additional function- These functions can be combined by connecting
ality including automatic power-off if temperatures the remote power switches to a remote console
exceed a threshold or automatic power cycling if server or by selecting a product that provides both
4 capabilities from a single unit. n
a network component becomes unresponsive (the
Energy use and data centres
Dannielle Furness, Editor, ECD Solutions

iStockphoto.com/DeepDarkness
J
ust as the industrial revolution did before There are three principle considerations for the
it, the information age has transformed our development and ongoing operations of a data centre,
world and the way we operate within it. Its the nebulous nature of which make future-looking
changed the way we access data and do decisions all the more difficult.
business; the way we communicate, educate and
consume information. It has given rise to entire
industries and countless changes to employment
Space
and careers; jobs have been created and others While the evolution of technology continues to shrink
rendered futile, often virtually overnight. the physical size of hardware and we live in the era
of virtual servers, theres no doubt the landscape has
Communication methods and the equipment we
changed considerably since the 1980s, when a 1
relied on 20 years ago, such as faxes and dial-up
GB hard drive was the size of a juke box. However,
modems, now seem as quaint as the quill and inkwell.
the sheer volume of data requiring storage ensures
The advent of mobile computing, the prevalence of
smartphones and a move to cloud computing in more that space still dictates the direction for design and
recent times have multiplied the effect and we now operations.
take for granted that digital files are permanent and Its hard to imagine total global capacity, but in
can be retrieved at any time, from any location, in February 2011, the University of Southern California
a matter of seconds. released research which calculated current worldwide
For data centre owners and operators, the chal- data storage at 295 exabytes, or 295 billion gigabytes.
lenge is to cope with the constantly changing face And it keeps growing; in a study conducted by IT
of the industry. Not only must they factor in the im- research company IDC in June of the same year, it
pact of continual shifts in customer expectation and was predicted that the world will generate 50 times
practices to adequately manage the mounting power current data production levels by 2020. Its all got to
consumption from this demand, but also crystal ball be stored somewhere. Its growing at such a rapid
into the future to ensure the projected life expectancy rate that in the not-too-distant future well hit a level
of the data centre (just under 20 years) is delivered that we havent even derived a term for yet ... but
5
and it meets commercial targets. thats another story.
 Power consumption hot and cool areas. Where possible, Google uses
water for cooling, rather than chillers.
Theres no denying data centres are power-hungry
beasts. Power to run the IT equipment itself, then The Open Compute Project, which has made public
power to run cooling and other environmental controls Facebooks so-called secret data centre recipes, is
and ancillaries like lighting. Power consumption in a a bid to encourage data centre development that is
data centre is often measured using PUE, or power more efficient from both a cost and power perspec-
usage effectiveness. PUE is the ratio of total power tive. After 12 months redesigning their server specs,
for the facility, including cooling, lighting etc, divided Facebook worked with manufacturers to achieve
by power utilised by the IT gear alone. a 38% increase in efficiency and a product they
maintain costs 24% less than the industry standard.
POWER USAGE EFFECTIVENESS They assert a PUE of 1.07 at the Prineville, Oregon
data centre. Google claims between 1.06 and 1.12
= Overall facility power/IT equipment across its centres, dependent on the interpretation
power. of total facility power usage (it claims it uses a more
Guidelines indicate an optimal PUE target of 1.0, stringent approach than others).
meaning that almost all power usage is consumed
by the IT hardware itself. Given the requirement for
Design
So, while not every data centre is on a par with Google,
cooling and environmental controls to ensure that
Facebook or Amazon, lessons can be learned from
ambient conditions are the most favourable for the IT
the way the way the big guys address problems.
equipment, its not uncommon to find a PUE closer
The same basic design principles apply and the
to 2.0. Not uncommon, but not ideal either.
problems they are facing today are the problems
of the future for smaller scale projects, particularly if
Cooling and environmental controls you consider the projected lifespan of a data centre.
Continual reliable operation is paramount in a data
To assist with the design process, professional
centre as any downtime can spell disaster. Hardware
organisations such as ASHRAE make a wealth of
is susceptible to overheating if adequate cooling and
information available to members including a com-
ventilation arent in place and even a few degrees
prehensive selection of publications specifically for
can make the difference between business as usual
the datacoms sector. These incorporate guides on
and catastrophic failure. If the installation is fortunate
best practice design for energy efficiency in data
enough to escape immediate failure, it can still suffer
centres, power trends and cooling applications and
delayed malfunction as fragile electronic componentry
real-time energy consumption measurements. See
can break down weeks after an overheating incident.
www.ashrae.org/bookstore for more.
Factor in loss of business, hardware replacement
and employee underutilisation during downtime and
Monitor, monitor, monitor
its easy to see that the costs soon add up and why
The importance of monitoring really cant be over-
operators are so keen to avoid it.
emphasised and, as the size and scope of data
How the big guys do it centres increases, visibility from a remote location
is imperative. With a simple monitoring system in
Some of the worlds bigger data centre operators
place, changes to environmental conditions that pose
including Google and Facebook, have been busy
a threat to system operation are identified before the
publishing information on their own centre energy-
crisis unfolds, via a web browser from any location.
efficiency initiatives, temperature control and other
cooling methods. Cynics might suggest that this Power usage monitoring is also useful and can
transparency is a PR exercise, but if theres lessons provide valuable design input for data centre upgrade
to be learned, why not take heed? projects in particular. Many solutions offer everything
from continual data logging and report generation,
Google suggests that most data centres are
which give a snapshot of the situation as it stands,
probably running cooler than they actually need to.
through to enabling corrective measures.
They cite the American Society of Heating, Refrig-
erating and Air-Conditioning Engineers (ASHRAE) Systems incorporating redundant power switching
and IT equipment manufacturers as expert opinion provide a reliable method of automatically switching
and suggest a slight temperature increase will not equipment to a backup power source, ensuring
only have no detrimental effect on equipment, but critical network devices are always up and running.
will deliver an immediate measurable energy saving.
There was once a time where the power draw of
The company also implements a design ethos a data centre would be the least of a contractors
comprising thermal modelling and airflow controls. The concerns. As long as the install went according to
modelling identifies potential data centre hotspots, plan, then job well done. These days, everyone on
so that equipment can be physically laid out in a the project has an interest in keeping power costs
fashion that delivers even temperatures across the down as the crossover between roles creates some
installation. Methods of airflow control that require no blurring of lines of responsibility. At the very least,
energy, such as plastic curtains and blanking panels, it makes sense to have an understanding of the
6
are utilised to ensure adequate segregation between factors that influence overall project power costs. n
How to remotely reboot
after a system lock-up

W
hen a piece of LAN/WAN, telecom or standard modems, only terminal emulation soft-
other control equipment has locked ware is required to dial the site and switch the
up and is no longer responding to power. Also, real-time communication with the
normal methods of communication, it reboot switch provides a response after each
is often necessary to perform a cold boot. After command has been accomplished.
the power has been cycled on and off, normal
Applications for the remote reboot switch can
communications via the network can resume.
range from common scenarios to more complex
This can be difficult at unmanned sites or when ones. Centrally controlled WAN environments
the problem occurs outside business hours. Even have a range of equipment, such as server
if a reboot is needed while someone is around, routers and dial-up equipment that frequently
you still have to hope that the employee is savvy lock-up and require a reboot. More specialised
enough to reboot the right device. scenarios involve satellite-controlled equipment
at communication towers, cellular towers or radio
For systems administrators, the ability to per-
equipment.
form a power cycle or remote reboot is also a
means of avoiding communication disasters. One The units can switch any AC powered device.
solution is a remote power reboot switch, which Heaters and air conditioners have been remotely
can be controlled by the systems administrator turned on and off at unmanned sites to protect
to ensure correct booting sequences in the event computer equipment.
of system failures.
Remote power reboot hardware is suitable
Such a remote reboot power switch is controlled for cluster management, where services are dis-
via ASCII commands. This means you can reboot tributed across a number of computer systems.
with a standard external async modem or over the
For applications that require high amperage,
TCP/IP network using a terminal server, comm.
heavy-duty/high-amperage reboot units have also
server or local server with terminal software.
been developed. The convenience of remote
ASCII commands sent to the reboot switch AC power control can be a welcome addition
can either query the current status, turn on/off to your current network management strategy,
or cycle (reboot) the AC power of any AC equip- and can also save time and expense of off-hours
ment attached to the switch. Since the reboot service calls. n
switch is controlled using these commands and

7
Image courtesy of Creative Commons
 The data centre of the
iStockphoto.com/Baris Simsek

next decade
Anthony Caruana, Editor, Technology Decisions
We are in the middle of the third great revolution of technology delivery. The
first was the mainframe era - where computing power was centralised and end-
user devices were unintelligent terminals. Then came the PC era. Marked by
massive increases in computing power, the pendulum swung completely with
client computers having more power and servers being relegated in importance.

W
e are now in the third wave. Many The way businesses look at delivering applications
services are centralised as data centres and other services must adapt. During the mainframe
have increased in computing power and client-server eras, the IT department had control
and capacity while end users enjoy a of the technology supply chain from infrastructure
massive variety in the types of equipment they can to applications. According to Trevor A Bunker, a
use and where they can use it. What does all this Vice President with CA Technologies, CIOs will be
mean if we are planning a data centre strategy that designing and managing data centres that bear little
will see us through the next decade? resemblance to those of today.

The nature of business and what CIOs need to The data centre of the future, from the CIOs view,
deliver to the business is changing at a pace that will not include the infrastructure. The infrastructure
is impossible to react to. CIOs and decision makers will be completely decoupled. I dont think that when
are working at a time where business cycles are CIOs think about the data centre that theyll even
contracting and change is accelerating. concern themselves with the infrastructure.
Robert Le Busque, Area VP Strategy and Develop- This begs the question - what is a data centre?
ment in Asia Pacific for Verizon explains: When the
curtain was falling on the Beijing Olympics the iPad In our view, the data centre is where applica-
didnt exist, the iPhone was only an infant and the tions, business communications and business logic
8
digital universe was five times smaller than it is today. reside. Typically, the data centre has also included
 had a significant impact on prices but theres little
doubt that large users of power will continue to see
the bottom line being impacted. Although theres no
universally agreed statement on what will happen to
power prices over the next decade, you can expect
your bills to increase by between 5% and 15% per
year over the next decade.

Googles answer to this is to build data centres

where theres access to cheap, reliable power. The
locations Google chooses for new data centres show
that access to cheap cooling, in order reduce power
costs, is as significant a decision as proximity to
communications.

Googles data centre in Hamina, Finland, is able

to take advantage of local seawater for its cooling
system. Phillip Sargeant of Gartner says: There are
a lot of providers of data centres today building data
centres in areas that they perhaps havent thought
about before. They want to make use of the charac-
teristics of the location. With cold locations they can
use outside air for cooling for example.

A critical measure of data centre power use is the

power usage effectiveness index. PUE is a measure
of how much power is used in a data centre for all
the elements in the data centre. The aim to achieve
a PUE of 1.0 - where all of the power being used by
the data centre goes directly to computing. When the
ratio rises above 1.0, the excess power is being used
for functions that support the computing operations.

Fujitsu recently upgraded a data centre in Noble

Park, a suburb of Melbourne, with the aim of reducing
the PUE to 1.7. The 6700 m2 data centre was built in
1988. Built to Tier III standards, it incorporates four
main data halls for cabinet and cage installations.
The company reports all greenhouse gas emissions
physical assets like servers, storage and networking produced by Noble Park, as well as all others in
equipment but those have always been in place to its Australian data centre network, to the National
serve the business. Greenhouse and Energy Reporting System.

Bunker believes that the physical infrastructure will

be less important as time goes on. Infrastructure
Location, location, location
Its interesting that when looking back at past re-
provides limited competitive advantage. Infrastructure
search what the issues were. In 2005, a Ciscos
is ubiquitous. Everyone has pretty much equal access.
advice focused on protection from hazards, easy
The real competitive advantage is going to emerge
accessibility and features that accommodate future
from how we use the IT services running on the in-
growth and change. A significant part of a research
frastructure regardless of where the infrastructure is.
paper described how to plan for natural disasters
Still, what about the data centre of 2025? What will and even listed earthquake statistics.
it look like? Will we still be constructing large rooms
Today, the location requirements are quite differ-
filled with hermetically sealed hallways and aisles of
ent. The accelerating density of computing power
blinking lights? Its hard to see a future where data
and rapidly increasing reliance on an always on
centres arent part of the picture.
infrastructure means that our expectations of data
According to Gartner Research VP Phillip R centres have changed. While the considerations
Sargeant, data centres are becoming far denser highlighted in Ciscos report are still important, there
with the amount of power thats required in a smaller are new things to consider for the data centre of the
space, the amount of heat that is being generated next decade and beyond.
and even the physical weight of equipment. All of
In order to minimise the operational costs asso-
this means that the choices CIOs need to make when
ciated with running a data centre, businesses may
building a data centre arent the same now as they
need to reconsider locations. In the past, it made
were even just five years ago.
sense to put the business and data centre close
There is not a single piece of data that suggests together. However, given that connections across
energy prices are going to fall in the foreseeable and between continents are getting faster and more
future. Every year we see the cost of electricity reliable, its possible to choose locations with access
9
rise. The federal governments carbon tax hasnt yet to cheaper power and better cooling.
 In the aftermath of the earthquakes that dev- meet demand. This cant be done without effective
astated Christchurch in early 2011, we toured the monitoring and automated power management
region and visited a new data centre operated by within the data centre.
Computer Concepts limited. CCLs facility avoided
being damaged - although the placard on the door What about the cloud?
telling us that the building had been checked was Theres no doubt that the decisions around what
a poignant reminder of the damage not far away to do about your businesss data centre needs
- but it also highlighted key considerations. CCLs will turn to the elephant in the room - the cloud.
facility was planning to secure its own water supply Past decisions were driven by different needs. As
for cooling so that it wasnt solely dependent on
iStockphoto.com/Baris Simsek

Trevor Bunker of CA Technologies puts it: Whether

power-hungry refrigeration. it was the purely centralised model years ago, then
Phillip Sargeant of Gartner highlighted to us that client-server, each of the evolutions weve done
some data centre operators are now looking to go for business applications has relied on one thing.
it alone when it comes to power as well. There are Thats LAN-speed network connections. Thats how
two or three data centres using trigeneration where we built our enterprise apps. We assumed that
people have their own power plants to power data they would run in the enterprise for the enterprise.
centres. Typically, some use natural gas to provide But those assumptions have been superseded.
power into their own data centres, he said. Bunker adds that Anyone whos thinking about
One of the challenges of any power genera- building a data centre - I would really have to
tion technology is the inherent loss that occurs. ask them why. Why would you make that capital
For example, when electricity is produced at a investment today? Is it really that strategic and
gas-powered power station, significant amounts of that valuable to your business? Is it a competitive
energy are lost in the form of heat. Trigeneration advantage for you? Many answer with frankly, it
seeks to avoid that by using natural gas to provide probably isnt. But its how weve done things in
electricity, heating and cooling. the past.

In addition, natural gas is a far more environ- Issues of data sovereignty, confidentiality, reli-
mentally friendly, and therefore cheaper, fuel than ability, connectivity and commercial arrangements
coal or many other alternatives. As carbon emis- dominate any discussion of cloud services. Its
sions become a greater impost on the bottom line, interesting that service providers are starting to take
being able to produce energy with lower carbon a more active role in our region with Rackspace
emissions can make a financial difference. opening a new data centre in Australia and making
specific mention of how it wont be subject to the
A recent trigeneration implementation by the Na-
Patriot Act although theres considerable debate
tional Australia Bank cost $6.5m but was expected
about the veracity of that statement.
to deliver $1m per annum in savings.
Both IDC and Gartner have recently published
research suggesting that a hybrid approach will be
Power management is a key a viable option. So, its likely that your data centre
Power management cannot be an afterthought in
in 2025 will have some local services and some
the next decades data centre. It requires as much,
either externally hosted or delivered as cloud ap-
if not more, planning and consideration than almost
plications. The physical footprint of your premises
any other aspect of the data centre.
will no longer bound your data centre.
Power management is more than just worrying
about the quality and reliability of supply and using
low energy devices. In order to execute an effective
What will your data centre look like in
power management strategy in the data centre,
the next decade and beyond?
It will be denser with more computing power per
managers need ensure that the supply to every
square metre than today. But it will also require
device is as reliable and cost effective as possible.
more power and generate more heat. Youll be a lot
For example, automatic transfer switches can smarter about where you build the data centre - if
automatically switch a device from the primary to a you build one at all - and youll probably start by
backup power source without interrupting operations. looking at the energy and carbon footprint as closely
This is done by constantly monitoring the power as the physical specifications of the equipment.
quality. Furthermore, the right power management
Youll consider making it either energy self-
equipment will also support remote management
sufficient or less dependent on power from the grid.
of devices by supporting remote power on, off and
system restarts in the event of a system becoming Where theres no competitive advantage or a
unreliable. clear cost benefit, youll probably use cloud services
where providers can deliver on your operational
Finally, smart power management strategies can
needs and energy management goals.
make it easy to detect which devices are busy and
adjust their power use depending on the workload. What is clear is that the days of companies
For example, research from Carnegie Mellon Univer- building large rooms with raised floors, expensive
sity suggests that some significant savings in power temperature management and large capital invest-
use can be achieved by powering off servers that ments are fading because the criteria for making
10 are not in use and then bringing them back up to the investment decisions are changing. n
 iStockphoto.com/Thomas Maier
Increasing uptime with improved
environmental monitoring
Workers and customers, empowered by smartphones and widely available Wi-
Fi services, want and are demanding 24x7 access to email, company network
resources and websites. And thanks to todays global marketplace, even small
companies must support round-the-clock activities.

U
nfortunately, IT system downtime remains contribute to or increase downtime and service
a problem for companies of all sizes. disruptions.
A 2010 eWEEK article reporting on an
Heat can be a killer. Extreme heat build-up can fry
industry study noted that North Ameri-
a server, knocking it offline and perhaps damaging
can businesses suffer an average of 10 hours of
it permanently. Even moderate heat build-up can
IT downtime annually. The article went on to note
have an impact. Equipment failure rate doubles for
that this downtime costs small companies about
every increase of 10C, according to studies done
$55,000 in revenue each year, while large com-
by the high-performance computing researchers at
panies lose about $1 million per year.
Los Alamos National Laboratory. Increased failure
To avoid the problems that can cause downtime, rate due to prolonged heating has also been noted
companies need to closely observe server room by the Uptime Institute and others.
environmental conditions and be alerted when
When it comes to monitoring temperature, it is
problems arise. This is an area where ITWatchDogs
not good enough simply to nail a thermostat to the
environmental monitoring solutions can help.
wall. Since the temperature can vary drastically
around different pieces of equipment, you should
Examining the causes of downtime consider placing separate temperature probes
11
Several data centre environmental factors can within individual racks or critical devices. That way,
 problems with a broken fan or an air-conditioning
failure will show up quickly. Similarly, you might be
able to identify a server that is overheating due to
it running an excessive workload.

To take nuances into account, ITWatchDogs

environmental monitors are designed for todays
crowded server rooms. They are small, ranging in
size from only 10 cm long up to the largest models
that are rack mountable at 48 cm/1U. The devices
can run off of existing electrical power outlets and
many support Power over Ethernet (PoE).

The monitors have built-in web servers and use

standard networking protocols, including TCP/IP and
HTTP. This allows server-room administrators and their
technical staff to monitor temperatures over an ether-
net network or remotely via the web from anywhere.

iStockphoto.com/Thomas Maier
The information is presented in a manner that allows
quick inspection of current temperatures, as well as
historical data to help spot heating pattern trends.
Finally, all ITWatchDogs environmental monitors are
capable of sending alerts via SNMP traps, email and
SMS messages. Some devices can also trigger an
external phone dialler to provide voice call alerts
for up to nine phone numbers, when predefined
thresholds are exceeded. A water main break in Texas took down the
computer systems in the Dallas County Records
Other server-room environments can cause down- Building. According to The Dallas Morning News,
time problems and need comparable monitoring and this [crippled] operations for almost the entire
alerting capabilities. county government.
Humidity is another major threat. The reason: Hu- Rains flooded a T-Mobile data centre in the
midity is the amount of water vapour in the air, and Pacific Northwest, taking down servers support-
too much water vapour can form condensation on ing the companys service activation portals and
electronic components, leading to electrical shorts. If websites.
the humidity is too low, there is an increased chance
Water is usually measured using a cable that is
of damage from electrostatic discharge. In either case,
run under an equipment rooms raised floor. When
uncontrolled humidity can severely damage critical
water comes in contact with the cable, an alarm is
server components, causing the server to crash
triggered.
and shutting down access to applications and data.
Proactive water monitoring should make use of
Unfortunately, humidity is one of the trickiest sensors capable of detecting the presence of water
environmental characteristics of a server room to over a large area so remedial action can be taken
measure and, as such, requires very close attention. before it shorts out equipment.
To measure humidity, most companies have A less frequent cause of downtime is fire and
focused on relative humidity. In fact, for years the smoke. In 2008, a fire destroyed 75 servers, routers
guidelines followed were based on recommendations and switches in a Green Bay data centre, accord-
of the American Society of Heating, Refrigerating and ing to Data Center Knowledge. Smaller fires and
Air-Conditioning Engineers (ASHRAE). The group sug- smoke from equipment or frayed wires can trigger
gested that the relative humidity for computer rooms fire-suppression systems which, while much better
be within the 40 to 55% range. However, because today at safeguarding equipment, can still cause
relative humidity varies with temperature, ASHRAE damage to IT equipment.
now recommends that data centres measure absolute
To detect fire and smoke requires more than tra-
humidity, expressed as the dewpoint (it should fall
ditional building smoke alarms. The problem is that
within 5.5 to 15C).
when they sense smoke there may be no one around
As was the case with temperature measurements, to hear it. Whats needed is an alarm that connects
humidity can vary significantly within a data centre. to web-enabled environmental monitors. In this way,
So sensors must be placed throughout the room the smoke alarm works as it normally does, but its
and server racks. alert can now be sent via an SNMP trap, email,
SMS and/or voice call to multiple IT staff members.
Water in a server room is never good news.
Whether the source is a leaking or burst pipe, or a ITWatchDogs environmental monitors come
flood, water can easily shut down an entire organisa- equipped with various onboard sensors along with
12 tion. Examples include: digital and analog inputs for external sensors, includ-
 remote power monitoring and switching capabilities
to any ITWatchDogs environment monitors supporting
a digital sensor port. The add-on accessory presents
real-time logging and graphing of voltage, amper-
age, real power, apparent power, power factor and
kilowatt-hour to provide trend analysis and power
metrics for future planning. The device enables us-
ers to set alarm thresholds for these measurements
and it can remotely reboot locked systems or control
system power via the secure user interface.

ITWatchDogs as your technology

partner
To increase IT system availability, organisations need
to take a proactive approach to monitoring the en-
vironmental conditions that contribute to downtime
and disruptions.

Certainly, for years IT equipment such as servers,

switches and storage devices have had temperature
and fan sensors, as well as software to send alerts
when temperatures rise or a fan fails. But in many
cases, these systems only notify you once a problem
is severe. Additionally, these monitors only give you
information about an individual device.
ing temperature, humidity, water, smoke and fire to Proactively monitoring conditions in the entire
name a few. The environmental monitors provide server room or data centre helps identify issues
a way to remotely monitor server room conditions, before they turn into a problem. This allows time to
view historical data to spot trends and receive alerts rectify matters before equipment deteriorates or fails.
when conditions exceed predefined thresholds. The
information provided by ITWatchDogs environmental ITWatchDogs offers a wide range of environmental
monitors can help a server room staff: monitors providing cost-effective ways for server-room
managers and their staff to proactively monitor their
notice changing conditions and take pre-emptive IT infrastructure and maintain system uptime. The
action to prevent downtime; products provide a quick and easy way to keep an
spot troublesome fluctuations and anomalies eye on remote conditions from a secure web interface
that might contribute to downtime; and receive alert notifications when specified alarm
receive alerts when conditions warrant immedi- thresholds are exceeded. The interface displays live
ate attention. video feeds and environmental measurements includ-
Furthermore, ITWatchDogs climate monitors can ing temperature, humidity, airflow, light, sound, power
be configured to display video feeds from up to four and water detection. The measurements are logged
IP network cameras. The interface provides a quick and graphed for viewing trend patterns. External
view of remote conditions along with environmental processes or applications can be automated on an
measurements when logged in. For a manager work- alarm trigger or remotely through the web interface
ing remotely or at home over the weekend, secure with units supporting output-relay control or with the
access to the interface is perfect to see whos in the Remote Power Manager X2.
server room and check whats going on from time
ITWatchDogs climate monitors use standard web
to time. Upon alarm, a quick glance can also help
server software to display their measurements and
determine if a trip to the facility is required or not.
camera feeds. All management and monitoring tools
And finally, when it comes to server room down- are accessible securely via ethernet or the internet;
time, the elephant in the room is power outages. no software installation is required. The monitors
Power outages are the leading cause of downtime. have SNMP agent software to integrate with popu-
Certainly, short outages can be covered with properly lar networking management tools and they support
configured UPS systems. However, in some cases, a SNMP v1, v2c, and v3.
UPS might further contribute to equipment failure if Most importantly, the ITWatchDogs line of prod-
it leaves servers running while the A/C remains off. ucts provides the proactive monitoring needed to
Naturally, if the power outage is longer-term - for maintain high availability in todays data centres and
instance, a severe winter storm tears down power equipment rooms.
lines - knowledge of the extent of the power failure For more information about ITWatchDogs, visit
is essential so that backup plans can be initiated. www.ITWatchDogs.com. n
For power monitoring, ITWatchDogs offers the Interworld Electronics is the Australian distributor of
Remote Power Manger X2 (RPM X2). This adds ITWatchDogs 13
 resources
from our sponsor

Founded in 1989, Interworld Electronics offers a full range of highly reliable industrial rack mount
and embedded computer solutions, data acquisition and communications hardware, audio visual
distribution systems, call centre recording, data centre power and environmental monitoring equipment.
Our products and solutions can control manufacturing processes, collect data, control transportation
systems, monitor power and increase data centre efficiencies.

Interworld Electronics is not simply a component re-seller; we focus on helping our customers meet
their demands. We work together with you as our partner in developing and delivering a total solution
for you or your clients. Our commitment to quality has attracted some of the best known public and
private sector companies in industries as diverse as telecommunications, mining, petrochemicals,
pharmaceuticals, defence, medical, transportation, call centres and data warehousing.

Interworld Electronics is an ISO 9001:2008 certified company.

Visit our website: http://www.ieci.com.au/

Physical Dangers Just as Important as Cyber-Threats
http://www.ieci.com.au/information/article.asp?Article_ID=863

Environmental Protection for Large Data Centres

http://www.ieci.com.au/information/article.asp?Article_ID=861

Power Management Products

http://www.ieci.com.au/products/class_index.asp?Class=9

Console Management Products

http://www.ieci.com.au/products/class_index.asp?Class=20

Environmental Monitoring Products

http://www.ieci.com.au/products/class_index.asp?Class=19