> Cover Story> Full Story
Ideal for e-services?
Dutta is Country Business Manager, Unix Servers
and Solutions, HP India. He is responsible for HP India's
Unix servers and solutions for Business Customer Sales
Organization (BCSO) division. He is presently looking
after HP9000 Unix server sales across India.
cost reduction, consolidation and high performance are the
current focus points for businesses offering e-services. The
Unix environment was designed to support these factors
Internet is an important way of life for us and is also critical
to businesses. Today a whole host of services are being offered
on the Net ranging from paying electricity bills to buying
movie tickets online you can even check the status of your
bank account at your office desk or from home. The benefits
of Web-based services (also called e-services) are simple:
speed, automation, cost-efficiency, and accessibility. Yet
to fully and cost effectively offer e-services on the Internet,
both service providers and enterprises need to ensure functionality
that is dependable and that can grow and adapt affordably
with company requirements. They should be able to deliver
highly available, highly reliable services, which are always
on 24 hrs a day, 7 days a week, 365 days a year to meet the
needs of their target customers.
Servers running Unix can address this requirement since, Unix
is the more reliable, flexible and scalable operating system
for any open platform. What's more, Unix is proven it has
been around for the last three decades. Unix systems have
always been a key component of the computing industry, but
the growth of e-commerce Internet sites and application service
providers, which rent access to these powerful machines, has
increased their value.
The Unix server market has seen an explosive growth over the
past 10 years. The top three vendors in the Unix server market
have all enjoyed an above average growth. The concept of Total
Cost of Ownership (TCO) has received a tremendous amount of
attention over recent years. Customers are becoming more savvy
and sophisticated in their analysis of systems. Ever increasing
budgetary pressures, combined with actual hands-on experience,
are causing more IT decision makers to look beyond a system's
initial price tag in their purchase evaluations. One area
where this is particularly true is with midrange servers,
which are often used as the heart of a company's network.
Although several TCO (Total Cost of Ownership) reports already
exist for these servers, they suffer from four limitations.
First, most of these reports are outdated at more than a year
old. Second, few, if any, take into account performance differences
between servers. Third, we are not aware of any study that
compares the TCO for Unix servers to those running proprietary
architectures. Lastly, none measure the TCO for newly introduced
systems. TCO is an important factor when purchasing RISC servers
since acquisition costs are generally less than one-third
of the server's TCO. This cost advantage is primarily due
to the low on-going operational expenses. Today high-end Unix
servers can easily match, and at times outperform even mainframe
class servers. So customers now have more choice when it comes
to replacing or buying systems to lower their TCO.
In order to maximize return on investment, managers factor
both the TCO and performance into their server purchase decisions.
Other factors such as software features, application availability,
and quality of service and support offered, also contribute
to the overall value of a server and should not be ignored.
In the past, TCO and performance were issues that have typically
been addressed separately.
Many leading businesses today are deploying a plethora of
information technologies seeking to improve competitiveness
and streamline operational costs. Concurrently, there is a
growing trend toward replacing yesterday's distributed computing
model with a more consolidated approach in order to seek greater
control, manageability, security and cost savings.
By consolidating applications, you can increase system utilization,
reduce resources needed to manage and maintain the systems,
and lower software costs and system footprint. This type of
environment may also be suited for partitioning technology
that is available on Unix servers.
It reduces cost of ownership and administration efforts, brings
higher service levels for users, boosts business flexibility
in terms of increase in demand, or changes in the environment.
Consolidating single and multiple application environments
reduces total cost of ownership and boosts operational efficiency.
The issue of how to increase existing server efficiency has
been rarely addressed; the server bottleneck caused by the
legacy network adapter or network interface card (NIC) has
not been fully recognized. So what is the most reasonable,
cost-effective, space-efficient solution to meet the need
for network speed?
When network performance and server response are unsatisfactory,
most people deploy more powerful servers with extra processors.
By doing so, vast amount of resources are consumed, including
personnel, space, real estate, and capital acquisition. But,
this does not necessarily speed up the network as expected.
The real catch is to design a network where all other loads
are distributed to dedicated networks, with specialized tasks
in mind, like the network for storage (SAN), for backups,
for clustering, etc.
For most server applications, pure MHz horsepower does not
yield a benefit commensurate with its cost. That means a costly,
faster CPU does not guarantee that the application transmission
speed can keep up with the claimed CPU clock speed or wire
speed. Other factors like the system bandwidth, memory access
speed, latency and cache technology make up for the performance
of the server. The evidence is clear organizations deploying
the technology can save money, improve server performance,
and reduce latency.
IT managers are justifiably concerned about reliability, performance,
scalability, ease of deployment and administration, affordability,
and lower TCO. Today's high-end Unix systems address all these
Server consolidation does not only result in a drastic reduction
of the TCO or money savings in terms of hardware and software,
personnel and real estate. It can also reduce the burden of
administering more servers and help increase productivity
across the enterprise. Server consolidation can provide IT
professionals an appropriate business solution using existing
server technology with emphasis on business savings.
So getting back to Unix, from a simple beginning as a personal
research project to an important role in operating systems
on a wide range of computer systems from desktop micros to
the largest mainframes class systems, Unix has and will have
a lot of impact. The strength of Unix is its portability across
multiple vendor hardware platforms, vendor independent networking,
and the strength of its application-programming interface,
along with scalability and reliability.
These benefits are so strong that the relatively weak end-user
interface has not slowed the adoption of Unix. End-users are
not direct beneficiaries of portability and application program
interface. However, end-users have already seen the dramatic
drop in the cost of computing when multiple vendors can provide
the same operating system and software solutions.