Here are two
paradigms:
1. Year: 2001
Organization: Navbharat Petroleum Corp.
Type of
organization: Petro PSU giant
Location: India
Application:
ourBAP.com, web-enabled ERP
Project Champion: Mahesh
Bose
The
Calcutta-based petro giant had taken on an ambitious, almost
impossible challenge of implementing an ERP across its country-wide
network of 10,000 offices and dealers in a short span of three
months. That was half the challenge. The other half was that till
now nobody in India had attempted to move completely to a
browser-based interface, across such a large number of nodes, with
limited bandwidth connectivity. And there was only one month left
now.
Bose reflected
on how the decision had been thrust on him.
The US-based global
giant, Gobil had decided to move into India and deploy a
fully-automated network within 12 months. Navbharat had to automate
in a record time or face oblivion. "Get us a solution that will
bring us on equal footing with Gobil. And you got three months to do
that", the management had demanded.
Bose sweated as
he reflected on what all could go wrong with no margin of time for
recovery. The RDBMS and ourBAP.com applications were still being
configured on the 100 NT server-cluster farm with load balancing
still being optimized. But he was confident.
His pilot of 100
nodes, well dispersed across the country, using the Citrix ICA
software had worked flawlessly. The 386, 486, PII and PIII systems
across India all had similar access and response times as the
cluster of 100 NT Terminal Servers at Calcutta. Only, instead of 100
clients he was scaling to 10,000.
Thirty days
later, all 10,000 clients accessed ourBAP.com flawlessly and without
a network glitch. Navbharat had automated and reengineered before
Gobil was halfway through its India plans. A press release by Gobil
15 days later announced that it was postponing its India plans due
to weak demand levels.
2. Year: 2001.
Organization: The home.
Type of organization: Medium-income
dwelling unit for family of four.
Location: Calcutta, India.
Application: 10th class arithmetic.
Project Champion: Mahesh
Bose.
Mahesh Bose was
not very happy. His 14-year-old son and 12-year-old daughter were
doing well in academics in their school. The school had made it
compulsory for all students to carry their own computing devices and
he had just purchased SmallPalms for them two years back. The
SmallPalm was good for the 8th and the 9th class, but now it could
not understand his son's 10th class Calculus. "What the hell, I will
have to keep buying one of these every year now," he thought grimly,
"Unless, I find another way."
His portable had
a wireless attachment. The Citrix ICA software on his portable
allowed him to run any application from the service provider,
including access to the internet. He thought to himself, "Suppose I
fix one of those attachments on the SmallPalm and put the Calculus
application on the service provider's server. Then I can keep
changing the application when my son reaches the 11th and 12th
class. I would not need to buy the MediumPalm and LargePalm
now".
So successful
was Bose's cost-saving idea that after being recognized as a role
model, the Parent Teachers Association protested to the management
of the school to make the same process convenient for all the
students. The school negotiated with an application service provider
and installed all the academic software on the server. The students
could now use any version of the Palm device to access applications
depending on which class they were in. The school added the
application access costs billed by the service provider to the
students' quarterly tuition bills. The cost of software access was
far cheaper than purchasing a new model of the Palm every year.
This is not
science fiction with a forward-looking window of a few years.
Neither are these forecasts made by technology consulting firms,
forecasting probable technology life cycles. This is the computing
paradigm of tomorrow, which has already started. The paradigm spells
'information appliances'.
On a more
practical note, at the Citrix iForum99 held in Orlando, Florida, Ed
Iacobucci, Chairman and Founder, Citrix systems, called it the start
of the age of digital independence. As he describes the paradigm,
"We are embarking on an age where an inflexible, complex and costly
technology industry that has catered to the 'digital elite' will
evolve to a new millenium industry that focuses on accessibility,
ubiquity and independence for all". Big words for a powerful but
simple meaning.
Client-server
computing has brought us till here, it's not going to get us much
further. From the vision of US-based International Data Corporation
(IDC), first there was the era of back office automation or
host-based computing associated with mainframes. This lasted from
the seventies to the eighties. Then came the era of front office
automation or client-server computing, which has extended into the
nineties. With the start of the next decade and the millenium, we
are faced with the advent of a new computing paradigm, the era of
automating the customer. The mainframe has given way to the PC and
the PC is now giving way to information appliances. The forecast
from IDC shows the sheer number of information appliances overtaking
PC shipments by mid 2000, with the growth rate of PC shipments
tapering off in subsequent years. As Iacobucci describes the huge
opportunity for the ubiquitous appliance, "There are a hundred more
people not connected to the internet than those connected to the
internet". But the 'elite' will continue to use the PC, now renamed
as the fat or thick PC, and piggybacking on the latent need for
ubiquitous information appliances in the enterprise is the rise of
the thin PC or thin client.
size=2>Server-based computing
face="Arial, Helvetica, sans-serif" size=2>
Ten years ago, when
Ed Iacobucci was working on developing OS/2 at IBM and Microsoft, he
figured they had got it all wrong. He felt that client-server
computing was too limiting, restrictive and only going to add cost
to the enterprise PC. He proposed server-based computing which was
the middle path between host-based and client-server computing. But
nobody listened or understood what he had to say at that time.
Iacobucci
created Citrix Systems in 1989, now a Florida-based $300 million
software organization. In 1999, Forbes named Citrix Systems as
'America's most dynamic software company' and Fortune ranked them #5
in the list of fastest growing companies. Iacobucci himself has
since been lauded with the Ernst & Youn Entrepreneur of the Year
award.
So what was
Iacobucci's server-based application model all about-that took a
decade to get recognized. Today, with the fast pace of computing
technologies, every PC in an enterprise is outdated from the day it
lands. Moreover, the average number of PCs in an enterprise are only
moving upwards and getting more dispersed. This is making version
control, application deployment and maintenance, and network
optimization and support, a CIOs' nightmare. Add to that,
heterogeneity of technologies and the fact that applications are
becoming increasingly browser-based and you get a full blown
calamity management in the enterprise.
Citrix now has
two products to enable a server-based computing environment in the
enterprise. Both are based on Windows NT Terminal Server. WinFrame
supports Windows NT 3.x and MetaFrame supports Windows NT 4.x. As
Nabeel Youakim, MD, Asia Pacific, Citrix Systems, describes the two
products, "WinFrame is for those guys who have decided to wait for
Windows 2000 rather than upgrade to NT 4.x." Both WinFrame and
MetaFrame work in the same manner by creating a shell for running
Windows applications. But that still leaves the client high and dry
on the other side waiting for user downloads. Unlike the network
computing model, there are no application downloads to the client in
the Citrix server application environments. The last piece in the
jigsaw solution is the Citrix ICA (Independent Computing
Architecture) software. The ICA software has three
components-server, network and client software. The ICA software on
the server separates application logic from the user interface and
transports it over standard network protocols and popular network
connections. Therefore, while the NT application is running in the
WinFrame or MetaFrame shell, ICA is transporting only screen
updates, mouse clicks and keystrokes up and down the network. That
is a lot less taxing on the network than a complete or partial
application download is in the network computing model. The
bandwidth required to support an ICA client is, therefore, typically
about 10-20Kbps.
Have we got it right? You really don't need a
lightning PC anymore to support an Office 2000 application. But you
do need good servers and even server farms. The cost of maintaining
centralized applications on a server farm is much lower than on the
client server or traditional model of computing. According to
Youakim, "A one way processor can support upto 30
clients".
size=2>Thin client
face="Arial, Helvetica, sans-serif" size=2>
The server-based
computing model from Citrix therefore advocates a small footprint
for the client in terms of processor and RAM with file retrievals
restricted to the server or network. A logical conclusion is that
286, 386 and 486 machines are also thin clients since they have a
relatively small footprint. And they can also be used to access
server-based applications since both WinFrame and MetaFrame support
early operating systems like DOS and Windows 3.x.
IDC describes a
thin client as any enterprise desktop technology that competes for
enterprise desktop real estate, primarily with PCs and traditional
terminals. In fact, IDC looks at thin clients as part of the
Terminal market. And Terminal devices have been around for some
time. The IDC Terminal taxonomy includes IBM 3270 terminals used
with mainframes and IBM 5250 terminals used with AS400 and the newer
System 3x mainframes. The IDC Terminal taxonomy also includes
ASCII/ANSI emulation terminals and enterprise thin
clients.
However, of
these four categories, the first three-IBM 3270, 5250 and
ASCII/ANSI-have been projected to be growing well below a negative
average growth of 10%, according to IDC. Only enterprise thin
clients show positive growth, a respectable 72%. Thin clients are
the preferred terminals of tomorrow with their share in the Terminal
market growing from 28% in 1998 to 94% in 2003. In 1998, thin
clients accounted for 689,000 units out of 1.5 million terminals
shipped. In 2003, thin clients will account for 6 million units out
of 6.3 million terminals shipped, according to IDC.
IDC also
admits that too much hair-splitting has been done in the enterprise
thin client product space, between Windows-based terminals and
Network computers. Windows-based terminals are shipped from OEM
partners of Citrix, like Wyse, Boundless and Neoware. Network
computers are shipped from OEM partners of the Java Virtual Machine,
like IBM, Hewlett-Packard, NCD/Tektronix and Sun.
In 1998,
Network computers accounted for 186,000 units and Windows-based
terminals accounted for 369,000 units. In 2003, Network computers
are expected to reach 1.3 million units at an average growth of 58%
and Windows-based terminals are expected to reach 2.4 million units
at an average growth of 81%. The Windows-based terminal is the
enterprise thin client of choice, according to IDC.
Are there any
significant differences between the Windows-based terminal or ICA
thin client and the Network computer or Java Virtual Machine (JVM)?
A clear statement that can be made is that they are not the same.
The ICA thin client is totally Windows NT-specific whereas the JVM
is server neutral. The ICA client does not download any part of the
application logic, whereas there are dynamic downloads for the JVM.
This results in a much smaller hardware and network footprint for
the ICA client in comparison to the JVM. The ICA client will not
need to be upgraded as frequently as the JVM. In the words of Mark
Templeton, President and CEO, Citrix Systems, "Independence was
always at the root of the ICA concept. ICA has been about the
freedom to overcome the complexities of computing." While the JVM
client, like the ICA client, was meant to be an alternative to the
expensive client server computing paradigm, it still has not rid
itself of lurking complexities.
size=2>The ASP business model
face="Arial, Helvetica, sans-serif" size=2>
Server-based
computing has another interesting spin-off. Thin clients in this
model would not have any local applications installed or deployed.
All applications and user licenses would relate to the usage of
applications off the server. Also, as connectivity and ubiquity
increases, the possibility of these applications being used off a
server outside the enterprise increases. An application service
provider would, therefore, necessarily need to offer connectivity as
well as application access. The user would pay for software
depending on extent of application access and usage. The traditional
model of an enterprise purchasing software license is therefore void
in tomorrow's model of ubiquitous access, and gives way to a utility
type model of pay-as-you-use. The ASP business model, while sounding
logical and common sense is fraught with difficulties mostly related
to changing the manner of business for vendors and their channel
partners.
size=2>The Indian perspective
face="Arial, Helvetica, sans-serif" size=2>
Within India, Citrix
solutions are marketed by Pune-based Datapro. There are a number of
enterprise users, who have tested the waters by implementing Citrix
solutions in one manner or the other. At Pune-based Thermax, Sudhir
Tandon, Divisional Manager, MIS, has implemented MetaFrame to run
Oracle, Lotus and Microsoft applications on his 386 and 486 PCs.
There are 35 licenses of ICA clients running off a Windows NT
Terminal Server. As Tandon describes it, "We saw a demonstration,
liked it and implemented it." However, Tandon is unable to extend
access of the ICA clients across the WAN since the leased line gives
unreliable performance. Adds Tandon, "Next on the agenda is to run
Baan off these clients."
At
Bangalore-based Silicon Automation Systems, Sunil Ghosal, Manager,
Systems Administration, has implemented the ICA client to replace
the X-windows Unix emulation. As Ghosal points out, "Silicon
Automation is primarily a Unix environment with some Microsoft
applications. The problem was to run these applications on the Sun
workstations." Now, with ICA clients on the Sun workstations,
Windows applications can be accessed whenever
necessary.
At New
Delhi-based HILTI, a construction consulting company, Ruchi Agarwal,
Manager MIS, is pilot testing the ICA client with Scala ERP
software. As Agarwal remarks, "We are satisfied with the performance
of running Scala from the ICA client." But so far, the pilot testing
has only extended across a single client.
At
Bangalore-based Infosys, all clients are running on Unix. That makes
access to Microsoft applications painful. Remarks Ilan Kumaran N,
Systems Engineeer, "We have about 40 ICA licenses for running
Microsoft applications from the single NT Server with Citrix
MetaFrame. But we seldom run more than 10-15 clients at a time,
since the response time becomes delayed after that."
While it appears
that Citrix WinFrame and MetaFrame have been tested in the Indian
enterprise with satisfactory results, the bogey is that the scale of
ICA or thin client deployment has been much too limited for the
cost-saving benefits of server-based computing to be realized. As
Pramod Nigam, Hindustan Lever, points out, "You need a large number
of these clients to be cost effective and with the poor
communication infrastructure, that is not very likely." Network
connectivity across the enterprise is a dampener in
India.
size=2>Looking ahead
face="Arial, Helvetica, sans-serif" size=2>
In circumspect,
Citrix applications in a heterogeneous environment are giving
immediate dividends to Indian enterprises. But server-based
computing and the concept of total cost of applications implies a
move-over to thinner clients on a reasonably large scale. As long as
Indian enterprises are hesitant of taking and overcoming this step,
the local computing paradigm is not going to change. With global
enterprises getting ready to take advantage of the lower costs of
server-based computing and thin clients, will India Inc again get
left behind in yet another technology overstep?