Data centers for business
Corporate data centers come of age
exponential growth and global organizations become prime focus areas, centralization
is the order of the day. The mantra for today's business is 'right information
at the right time', and the corporate data center plays a pivotal role. This
is why we advocate that, the time is right for data center-ic businesses. by
Anil Patrick R
In today's organizations, data centers play a driving role in business advancement.
As is visible in many Indian organizations, data centers have evolved from being
looked at, as mere 'technological fancies' to actual business needs.
Data centers, which initially made their major appearance
in India circa mid 1990s have matured into essential business drivers. The main
reason behind this has been that companies have been able to identify the benefits
of aligning technological advantages with business needs.
Business heads have graduated to a stage where they
know the overall IT vision. They are only concerned with the end result, not
the exact IT implementation details, said Satish Pendse, Head - IT, Kuoni
Combined with the need for real-time data access and regulatory
compliances, the time is ripe for more organizations to opt for data centers.
The earlier argument of poor connectivity going against centralized infrastructure
in India is no longer valid. With factors like redundant connectivity from multiple
service providers, easily available technical expertise, and decreasing hardware/implementation
costs, the time is ripe to deploy a corporate data center.
The data center as such is not a business need, but
the benefits given by the data center are the business benefits. The end results
that the business expects usually need a data center to fulfil, said Pendse.
Sagar Sule, VP - Technical & Operations, Cyquator Technologies
also attributes the contribution of technology to business growth. Nowadays,
technology has become synonymous with profitability and ROI. As data centers
are very '3M' (Man-Machine-Money) intensive, their perceived technology need
has transformed into a business need, said Sule.
On the technology front, the benefits of having a data
center far outweigh the traditional distributed approach. But before going into
those, we need to have a look at the business benefits first.
'Data center'ic business strengths
At first glance, cost and infrastructure management benefits
might seem to be the only advantages of having data centers. While costs are
important, benefits like lower business risks and higher user productivity levels
that data centers present are also significant.
Today's organizations (especially those like BFSI, telecom,
manufacturing, call centers, and software) require customer/supplier interaction
over the organizational network. Even the slightest downtime or performance
lag can cause irreparable loss of valuable customers, reputation, and business
damage. It can also be detrimental to employee productivity levels, especially
for businesses dealing with outsourcing.
Distributed infrastructure models can be fatal for such
businesses. These models usually produce multiple information silos, which are
highly failure prone, hard to manage, harder to detect/recover from failure,
and difficult to scale. With data volumes increasing tremendously, these models
can develop performance bottlenecks and backup/recovery nightmares in case things
go wrong. Such architectures can also be detrimental to plans of voice traffic
over the network.
This is why it is beneficial for organizations to deploy
data centers. These operating models ensure a centralized environment with standardized
processes offering lesser operational risks. This is made possible due to the
higher manageability/administration and lesser point of failures.
These business advantages alone ought to justify the need
for a data center-based approach, but the cost/management perspective is similarly
interesting. Refer to Data center-ic business models to see some
of the major benefit reapers of data center driven business.
The cost advantage
When sized and structured according to the business needs,
data centers offer cost benefits on fronts like hardware, IT capital expenditure,
and management/administration. In the case of a data center, increasing cost
pressures can be dealt with more effectively than is possible with a distributed
The cost of server and storage consolidation is lesser
than that of a distributed environment, and provides better cost-efficiency.
For example, the cost of providing redundancy in 50 locations is much more than
that required for a single location. Usually, organizations provide around 33
percent of redundancy per location. When put together, the costs will be much
greater than that required for a single data center location.
Human resource savings in terms of the trained personnel
who need to be present at each location also come down with the deployment of
a data center. Management of the team becomes easier.
With all the major components in a single environment,
securing the data center environment is simpler.
Disaster Recovery (DR) with centralized backup/recovery
is also simplified with this infrastructure. Downtimes come down significantly
with this approach. And outsourcing the data center's management becomes greatly
simplified when there is only a single location.
Organizations have two options when they decide to implement
a corporate data center. They can either host it with an external Internet Data
Center (IDC) or build the data center themselves. Both approaches have their
pros and cons.
Indian organizations have a major advantage over organizations
around the world since they will be effectively 'leap frogging' over the evolutionary
period of data center history.
The co-location option
If an organization has a small IT team, and intends to
focus more on the core business than supplementary data center activities, hosting
at an external IDC is more viable. In such cases, there is no need to worry
about factors like power supply, server space, building infrastructure maintenance,
and the security required for a data center environment.
Co-location is also an attractive option for organizations
that want to make minimal investments in hardware. Large data centers hire out
their equipment, infrastructure, and space. All that is required from the organization's
side are the employees to manage the infrastructure. All the data and applications
are owned by the organization. Only the hosting services are hired.
Many IDCs also offer management services. Another area
for co-location is in hosting specific applications where the organization is
waiting for its infrastructure to be created.
We will examine the various co-location options available
in detail in the next story External help with a data center.
In-house data centers
In the case of large organizations running multiple applications,
co-location is not always the best option. The biggest drawback of co-location
is that the operating model is more suited for running a limited number of applications.
With increasing applications, it is always best to use the organization's own
data center to have the best performance, management, security, and accessibility.
The security in a common environment is always a big issue.
It is possible to impose your own security environment in a co-located
environment. However, in the long term, the reliability of this depends on the
size of the organization. A large bank needs to have its own control on these
procedures, said V. K. Ramani, President (IT), UTI Bank. In addition to
this, many regulatory authorities do not permit termination of their connections
into a common environment like an IDC.
Next is the performance aspect when too many connections
are involved. When you are co-locating certain facilities like processing,
the leased lines become a bottle neck. Too many hops and delays result in poor
performance and can lead to unhappy clients. This is the primary reason why
we looked at hosting our servers in-house, said Manoj Chandiramani, VP
- Networks, Refco-Sify Securities India.
Many organizations prefer to handle application management
themselves. The prime reason is that support staff in co-located facilities
look at different customers across varying verticals, platforms, and applications.
Support staff fails to have expertise and are usually
unable to pinpoint a problem immediately in this scenario. What happens with
in-house management is that the team is working only on the organization's setup.
They know their job and servers better than anyone else, said Chandiramani.
When designing an in-house data center, it is necessary
to have a clear idea of the present and future business needs to be derived
from it. This is where the role of the business comes into the planning stage.
Unless the business is clear about the direction of growth,
it is not possible to properly plan and size the data center. Involvement of
the business also helps determine the financial feasibility of the project-what
is possible and what is not.
Once these details are sorted out, it is time to chalk
out the actual technology involved. The key parameters to keep in mind during
this stage are scalability, availability, manageability, security, reliability,
and supportability. See Data center standards for more information
on common data center standards and best practices that are useful at this stage.
During design, it is crucial to follow data center standards
in terms of physical infrastructure, security, access control, network control,
and administration. Many organizations involve vendors and consultants during
these stages. If the organization does not have prior experience in implementing
data centers this involvement can be beneficial. However, be prepared to closely
monitor the activities.
If you are using a consultant, you have to be careful
about the materials that go into the data center, particularly in terms of fire
protection, access controls, alarm systems, and total control of movement &
materials into the system, said Ramani.
When planning a data center enough advance planning is
needed. It should be possible to scale up the data center when required by the
business requirements. A common practice is to size the data center so that
it can cater to future business requirements for at least three years. Next
comes choosing the location. It should have expansion capability for future
growth in terms of parameters like available space, costs and infrastructure.
The available space consideration should also provide for future technology
Considering the fast rate of technology obsolescence, a
complete physical rearrangement to accommodate new technology might be required
in future. Infrastructure parameters to consider are those like redundant connectivity
links from multiple service providers, dependable redundant power supply, backup
power, and physical security. Other parameters include availability of skilled
manpower and proximity to the IT team/most needed location.
It is also necessary to have necessary business continuity,
disaster recovery and back up plans in place. See story Prepare to recover
for more details.
Data center policies
The company must have proper documentation and enforcement
of the data center policies. It is easy to draw up laborious systems and processes.
But it is quite another thing to actually enforce these in the long run. This
is the only way to get the optimum results and ROI from the data center investment.
With inputs from Soutiman Das Gupta.
Anil Patrick R can be reached at email@example.com
|When it comes to data center-driven businesses, telecom,
and BFSI segments lead the field. The main reasons for this are the data
heavy nature and interaction with end users that are typical to both industries.
These requirements call for a centralized data center-based architecture.
In industries where end users are involved, there is a
large number of data touch points. This is why adoption of data centers
in the services industry is happening on a very large scale. Today, services
industry segments like airlines, hotels, and call centers operate out
of data centers. These industries cannot afford even the slightest amount
of downtime. Data center-based operating models help them get optimum
performance and seamless operation out of their infrastructure.
Another area that is going to start adopting data center-based operations
is the manufacturing segment. With enterprise applications like ERP and
CRM churning out huge volumes of data, data management is required even
for function-centric distributed applications.
- Enterprise data distributed across the organization.
- Windows or NetWare as a file serving OS or
one of more than 100 Unix
- variants in a scientific or academic environment.
- Systems with attached hard drives of less
than 25 GB the big file
- systems supported less than 256 GB of data
- Single function servers.
- Backups for data integrity
run locally to tape using basic systems utilities.
- Consolidated nature of data.
- Clustered Unix and Windows servers surrounding
- NAS and SAN attached storage is common.
- Storage management utilities are widely used.
- Addition of tape robotics
to the network to virtualize tape capacity and drive usage.
- More similar to the switching network than
any currently available server architecture.
- New processors, memory extensions and additional
storage devices can be dynamically added to the processor cluster complex
- Intelligent disk arrays superceded by intelligent
storage network fabric, storage consolidation replaced by centralized
|Setting up a data center alone is not enough. It is also
necessary to have sufficient backup, DR, and Business Continuity (BC) measures.
These will help keep the organization prepared in case things go wrong.
While backup is a comparatively easy proposition, DR is
a completely different ballgame. There are more factors involved in DR
than just database replication at a remote location. For example, the
network must have an alternate path to continue operations as soon as
possible. This is why DR is more about the ability of a data center to
be back in operation with all the applications back in place.
More important than maintaining data integrity is knowledge
of the minimum interval for resuming operations. It should be possible
to guarantee the timeframe in which the percentage of desired efficiency
could be brought back. This can be achieved only by having the proper
systems for this, and sufficiently skilled personnel. The recovery team
has to be conversant with the processes involved in bringing back the
The next issue is about how quickly it is possible
to get back to the original host and ensuring that it's up. These will
define the provided service levels and this is where the organization's
business continuity plan steps in. DR is an event-based term that is just
a subsection of the BC plan.
All the elements of the DR plan have to be defined and
implemented according to the BC plan. The mistake most organizations make
is in relegating DR and BC plans to the IT department. A business cannot
recover from just the IT team's efforts. Every section of the organization
is concerned with BC.
So it is imperative that every department is geared up
with the appropriate plans. This initiative is absolutely necessary when
planning for total centralization of the data center and operations.
|In India there are many best practices and standards
used widely. Best practices followed include best practices for power, data
cabling, dust & humidity levels, and heat convection.
IT Infrastructure Library (ITIL) and IT Service Management
(ITSM) standards are becoming increasingly popular with data centers.
BS 7799 is a popular security standard being widely used. Many data centers
do use vendor specific standards published like the SunTone Certification
and Microsoft Operations Framework.
The DCML consortium (dcml.org) led by EDS is working on
the Data Center Markup Language (DCML). DCML organization claims that
the language enables IT organizations to benefit from the utility computing
vision of lower costs, increased agility, and improved service levels.
The data center language standard is expected to facilitate automation
of data center environments by improving interoperability among cross-vendor