When Data Spells Money

Data centers are definitely the flavor of the season.

Speaking of flavor, the best analogy I have heard about a data center is that
to a dosa, the favorite Indian crepe. Most people like it; some might even be
able to rustle one up. But it is not pertinent that only those who can cook it
can eat it.

“When a customer goes to a restaurant and asks for a dosa he has a fair idea
about what he will be served, how it will look and what it will taste like. He
is not bothered about the ratio of water to the batter, the heat at which it has
to be poured on a pan or the right time to take it off the heat. He wants a meal
which has more or less a universal standard, which is exactly what a data center
customer is looking for-a predictable environment for running his business
applica­tions and storing his data,” explained Durgadutt Nedungadi, Director
Marketing and Alliances, Hewlett Packard (India).

Little wonder then that most vendors are falling all over themselves to
showcase their data center-relevant products and are busy evangelizing how their
solutions are best suited for customers in India, banking on dishing out this
very predictable environment. Any solution provider will tell you that setting
up a data center is no mean feat. There are challenges galore that are involved.
And the biggest challenge is about increasing the center's performance while
bringing down its costs.

This is a very generic concern but it can be broken down into specific
subsets.

Challenge #1: Construction

The construction of a data center is a science by itself. There is a classic
case where a company set up a data center in Ireland where there was ample sun,
wind and water. The data center was powered by solar, hydro and wind energy, and
the natural airflow and availability of water helped in the cooling of the
systems.

“Building a data center itself is a big challenge, as there are certain norms
that need to be followed and it's a pretty scientific job,” said Karthik Ramarao,
Technology Team, Systems Practice, Sun Microsystems. According to him there are
a variety of challenges faced during the construction of the data center,
starting from location of the landscape.

As in the hospitality business when it comes to data centers too what will
determine its success is its location, which is also the foremost challenge.
According to a survey, by 2010 more than half the existing data centers will
have to relocate, while more than 40 percent will be replaced within the next 10
years.

Besides having good interior features, the facility must also have a good
outer space. Time and cost are also some of the important challenge faced during
the construction of a data center.

“It takes several phases for completing a data center, which imparts pressure
on the investments made,” Ramarao added. According to him the selection of
cables do matter and this is where most solution providers (SPs) make a big
mistake.

Bhupendra Rane, Country Sales Manager, Tyco Electronics pointed out that most
data centers are stuck in the dark ages where they use conventional fiber
technologies and therefore have patch cord management issues and there is a need
to manually intervene for making any changes.

Modular data centers that are pre-built with racks and necessary power and
cooling solutions in place is the solution Sun has come out with. “In Sun we
have a solution for the constraints in construction of data centers in the form
of modular data center, which is a pre built data center that could be deployed
anywhere, anytime. A data center of 6,000sq ft could be easily fit into this
modular data center, which could save a lot of time and maintenance cost for
data center administrators,” Ramarao informed.

If you are setting up a hosting data center then you could factor the fact
that some of your clients would want to have their servers in cages with better
security so that their IT personnel can work on consoles for some application or
software related issues. This also means that you would have to create some kind
of seating arrange­ments for these people within the data center as well.

Challenge #2: Incorrect choice of equipment

Data centers that were set up some years ago have a peculiar problem. Most
of them were put together with the best of breed of equipment, but nonetheless
their functionality and manage­ment can be cumbersome. This is because just
having the best products does not mean they will work best. What is needed is
equipment that will talk with each other or are interoperable.

“Most of the mistakes are made starting from cables, due to cost factor,
where the multimode cables and CAT cables are interchanged accor­ding to the
financial strength.

Lot of such factors need to be executed carefully else everything would
finally affect the downtime of the data center,” Ramarao noted.

Agreeing with him, Nedungadi said that he had actually taken a look below the
floor in data centers and realized that his son's bedroom was better organized.
“A maze of wires runs across the entire grid. If there is scarcely any space for
a rat to pass through, how do you expect hot and cool air to circulate freely
without affecting the wires?” he questioned.

This is why it is important to have a facilities planning done at the
strategy level even before contemplating setting up a data center. SPs should
ask questions like what are my clients' IT need going to be over the next five
years, how much power will be required to run this, what is the nature of
hardware and software that will be needed, etc. It is only after getting answers
to these questions should the SP move to strategy setting for the data center to
designing it to rolling out the operations.

Challenge #3: Power and cooling

Power consumption and cooling are the primary challenges faced by any data
center with power and utility bills rocking the data center managers. The power
bills amount for 60 percent of the data center operational cost.

This is because the design of data center predominantly are in a manner,
which consumes huge amount of power and radiates as much of heat energy it can
and that in turn will have an impact on the power due to the amount that is used
to cool the facilities. According to one survey, over the next five years, power
failures and limits on power availability will halt data center operations in
more than 90 percent companies.

According to Rajesh Saha, VP-Enterprise Business, STG, IBM India/South Asia,
the biggest challenge in setting up and managing the data center is emission of
carbon dioxide or CO2. Worldwide, data centers are reported to consume 40
terawatt hours of electricity each year, producing an estimated 17.2 billion
tonnes of CO2 emissions.

“With the global tempera­tures on the rise, many environmentalists are
calling for significant reduction in the generation of CO2 from fossil fuels.
Commercial electricity consumption is a major factor in rising atmospheric CO2
levels. This is a concern for all and needs to be addresses by all parties
involved,” Saha said.

IT accounts for two percent of all global carbon emissions, placing it on a
par with the aviation sector according to a Gartner Group report. The research
agency released a report that highlighted that in an estimated distribution of
global CO2 emissions from IT devices, 40 percent comes from PCs, 23 percent from
servers, 15 percent from fixed line telecom, nine percent from mobile telecom,
seven percent from LAN equipment and the balance six percent from printers.

There is a clear need for SPs to educate themselves and their clients about
the cooling require­ments within a data center. There is no need to refrigerate
the equipment. Networking and storage equipment servers typically have a cooling
require­ment of 22 to 26 degrees. So the ideal temperature of a data center
should be around 24 degrees to ensure that the equipment does not malfunction
and this will also bring down the cost of cooling leading to better price
performance of the data center.

All the latest data centers have hot and cool aisles. In the former the backs
of the servers are placed facing each other and the heat that is generated is
then sucked through air vents overhead or underground. In the cool aisles the
machines are fed with cold air which helps them function optimally.

Said Sharad Sanghi, CEO of Netmagic Solutions, “It is therefore important to
organize airflows efficiently by placing racks properly and deciding how much
and what kind of cooling is needed.” This approach also helps in dissipating
heat rather than isolating it in one area. Additionally with the creation of hot
and cold aisles it is easier to regulate the temperature of one area than bring
down or increase the temperature of the entire floor in a data center.

Challenge #4: Complexity in services

For Ganesh Mahabala, Regional Director, VMware India and SAARC, the key
challenge that a SP faces in the maintenance of the data center is the
increasing complexity in services. The complexity in the applications and
technology are constantly increasing and the industry is moving away from its
classic way of working towards a further complex environment such as web-based
application, software as a service (SaaS) model and service oriented
architecture (SOA).

Power consumption and cooling are the primary challenges faced by any data center with power and utility bills being the primary concerns of data center managers

“The customer's high-end requirements are further making the scenario worse
for SPs to manage the existing resources, as new resources keep coming in a
multiplied manner,” noted Mahabala. He added that these increasing complexities
in some way are reducing the efficiency of data centers.

Ramarao also feels that innovations keep happening in the area of data
centers in order to save cost, time and power. “An enormous amount of time is
spent in the R&D activities, were we are trying to bring out solutions that
would bring down the cost of power and the down time of the data centers. The
processors play a vital role in heat generation and considering this Sun has
launched Niagara T 2 processor, which performs enormous tasks by generating less
heat. Moreover, technologies like virtualization can save power by increasing
the efficiency of the servers, as it enables an effective usage of computation
power,” claimed Ramarao.

It is also important that SPs should move away from providing silos of
solutions to a single unified service oriented option to customers through
consolidation. This is because the customer stickiness to a data center is
usually high as they would not want to move their entire infrastructure to
another data facility unless absolutely necessary.

Said Nedungadi, “Don't merely focus on having servers for one division of the
business. Get the client to let you create a pool of infrastructure.” Of course
this also makes the SP accoun­table for the quality of service provided for the
entire business environment rather than just one business domain. But that is
the way to capture clients holistically.

Challenge #5: Storage issues

Like power and energy, storage is a key issue faced by SPs during the
management of data centers. More than ever, storage has become an important
element of IT infrastructure. The main driver for this shift is that of a
balanced performance requirement.

During earlier days, storage used to be an afterthought and was not a
significant factor in influencing data center performance. With the
ever-increasing storage capacities, corporations are paying a lot of attention
to ensuring that storage performance and capabilities complement that
application environment.

Saha of IBM informed that from a storage perspective, the dynamic business
landscape causes three essential issues that vendor are trying to tackle. “The
primary issue is the variety of data available. Here companies are dealing with
various applications ranging from databases to mails to rich content etc,” he
said. Storage devices are expected to exhibit different characteristics, based
on the workload and data type.

The second issue is the volume of data. It is an undispu­table fact that the
data being accumulated by companies has increased dramatically. Thirdly the
velocity of change is also a huge issue, where the rapid change in the business
forces IT to adapt quickly. “This causes IT solutions to be rolled out without
adequate validations resulting in poorer user experience,” Saha averred.

Unanimously vendors suggest storage virtualization as the solution for the
increasing woes in the storage management area. In storage virtualization, data
can be stored in a tiered model, which increases the efficiency of servers to a
great extent. As a result of all these storage, server and network
virtualization the administration work becomes lesser.

“Deployment of robust applications gives unprece­dented flexibility and
availability to meet the increasing demands of the business. Similarly blade
servers are a solution for increasing storage issues in the data center, where
it consumes less power, needs less server room cooling and most importantly it
has an impact on cost savings. Vitalizing the total environment would be the
solution for managing data centers effectively,” Mahabala informed. But the
virtualization should be aimed more at consolidation than replication because if
you replicate physical data on a server into a virtual environment it will only
be self-defeating.

Industry veterans concur that unified storage is the best way to manage the
data center without heavy investment. People are talking more about the unified
storage, using which it is possible to run and manage files and applications
from a single device and can be accessed through NAS, SAN, ISCSI using
virtualized storage.

Here, the user doesn't know where the data resides but can access it. The
data can be moved seamlessly among the storage technologies. “Also, by using
data deduplication one can compress existing data into one tenth of its size,
without compromising on the quality,” Mahabala informed.

Saha of IBM suggested infrastructure management platforms for the effective
management of storage in data centers. “This pivots storage monitoring and
envelopes other infrastructure elements such as tapes, network fabric and file
systems. With the level of growth that we are experiencing in this region,
virtualization platforms provide much desired flexibility into the
infrastructure,” he said.

“Virtualization not only saves on the servers but also on the networking,
storage and cabling. It will enable a user to utilize every rack space available
in the data center,” Mahabala claimed. According to a Forrester Research survey
Windows servers typically run at only eight to 12 percent of their full
capacity, and UNIX servers at only 25 to 30 percent, which shows just how
underutilized the equipment are in a live data center.

Challenge #6: Management

So you have your client's data in a center, which is miles away from where
the client is based. How do you manage it? Remote management is the answer.
Remote management through KVM systems reduces the need for extra keyboards,
monitors and mouse devices. It also offers data center managers remote access to
their servers for monitoring performance and power consumption of their
hardware. “KVM remote manage­­ment tools also enable data center managers to
make server performance modifica­tions remotely over the Internet,” said Kevin
Chen, CEO of Aten International.

With remote KVM products there is no reason for the client's IT administrator
or the SP to physically travel to data centers to make any application changes
in person. This is critical for disaster recovery efforts since most companies
maintain two data centers spread across considerable distances to mitigate
security breaches. Also, KVMs do not restrict remote communications to a few
users, especially when an entire team needs to diagnose major network issues.

Besides this, proper use of KVM tools significantly reduces the physical
square footage space and associated power required to run a data center. This is
achieved by optimizing rack space and eliminating the clutter, power and cost
required to supply additional hardware.

“On average, a monitor consumes 50W per hour. Multiply that number by
24x7x365 consumption and you have a huge power bill staring you in the face.
Now, by using KVM, the number of monitors running in a typical data center
decreases and the environ­mental savings quickly add up,” added Chen.

LCD KVM switches that typically slideaway or hideaway within a rack actually
cut the power consumption to 28W and 30W-including the LCD monitor. In addition
to energy savings, LCD KVM switches also reduce data center real estate since
these products tend to only take up 1U of rack space versus two separate units
for the KVM and LCD console.

Challenge #7: Open source or proprietary

There are many purists who would like to believe that a data center should
run entirely on proprietary products-be they hardware or software. But at a time
when bring down data center operation costs is one of the key concerns of SPs
there is an increasing breed willing to consider open source.

Rahulkrishna Gupta, Techni­cal Sales Specialist, Novell India stated,
“Histori­cally, having a proprietary platform means that customers are locked
into it and need to upgrade it by buying expensive versions and hardware. Also,
proprietary brings with it inter-operability issues.”

This is why companies like Red Hat and Novell are pushing for increasing
adoption of open source in data centers. They argue that open source saves 40
percent of hardware cost over Unix and around 34 percent of software cost over
proprietary platforms like Microsoft.

“There has been a CAGR increase of 15 percent of Linux adoption in data
centers between 2005 to 2010. Besides, according to a survey, 69 percent of data
centers have deployed Linux in some form in data centers. This shows that
customers are willing to opt for open source and quite a few of these
deployments have been in mission critical applications as well,” noted Gupta.

He added that with a third generation Linux operating system, there is a
policy based automation of tasks which frees up the personnel managing the data
center who can then be used for other business purposes. “There is also an
automated lifecycle management of both physical and virtual resources,” Gupta
said.

Challenge #8: Business continuity and disaster recovery

Recently, a leading telecom service provider's data center in India caught
fire and the connectivity to almost 60 percent of its subscribers was affected.
This can happen to any company, which is why it is important that its data
center should have a well planned business continuity and disaster recovery plan
in place.

“Business continuity and availability is about ensuring that an organization
can continue to meet its obligations and commitments to a level of operating
capability that is acceptable to that organization, its customers/users and the
regulations or best practices it must abide by,” defined Jane Rushton-Young,
Regional Manager-Data Center Services and BC&A Technology Services, HP (India).

She added that disaster recovery or IT outages in the 70s and early 80s has
evolved into business continuity in the late 90s. Today it implies a more formal
set of commitments by businesses in terms of availa­bility of both IT systems to
meet the faster pace of business in an ever increasingly competitive world as
well as the availability of various business functions that businesses commit to
as part of customer contracts, supply chain agreements, regulatory needs,
competitive imperatives to retain its market position, industry standing and
maybe very survival.

At a recent industry event Anand Naik, Director-Systems Engineering, Symantec
India and SAARC noted that in 1990 business continuity scenarios were limited to
technology or component failure. Soon businesses realized the need for business
recovery and not just IT recovery. “Over the last 10 years, world events,
Internet explosions and increasing dependence of businesses on IT has reduced
the recovery point objective (RPO) and recovery time objective (RTO)
requirements for various organizations. For most critical applications and
business functions, this has become real time. Due to this change, business
continuity needs to be now looked at proactively,” he noted.

Naik informed that by opting for this approach the data center is protected
against and can prevent data loss and downtime besides avoiding outages via
proactively monitoring processes. “But if your facility has a business
continuity problem then the best way to fix it is by identifying vulnerable
systems, technology and process gaps, application failures and data loss,” he
added.

Rushton-Young on the other hand talked about the need to be resilient when it
comes to business continuity and IT outages. “As IT downtime extends from
minutes to days, there is usually an impact on productivity and revenue-on what
scale is something you need to be able to identify,” she advised. “With a
holistic approach to reducing downtime and a good understanding of what the
impact is across different parts of your business you can start to look at how
resilient you are and how resilient you need to be,” added Rushton-Young.

Conclusion

These are just some of the glaring challenges that SPs setting up a data
center, be it a captive one for their customers or a hosted one, will face.
There are others that exist as well. But touching upon each of them will take up
an entire volume of pages. This shows that there is a lot of scope for
innovation when it comes to data centers. And innovation is what SPs in India
are best known for.

VINITA BHATIA

vinitavs@cybermedia.co.in

(With inputs from Sethuraman NR)