Open source, new tools and technologies, and offshoring
make building, rather than buying, an application an increasingly
viable option.
Five years ago, it seemed as if the debate about whether to
build or buy applications was over, with the forces for buying
having won the day. Today, conventional wisdom states that the
entire market for packaged applications will consolidate to a few
mainstream suppliers, such as SAP or Oracle.
This is an over-simplistic and misguided view. There are many
markets and submarkets where the option to build is prevalent. This
does not mean that companies should consider building a word
processor or database. Nor does it mean that companies below a
certain size should only consider building their own software.
Rather it means that the build-versus-buy pendulum has swung back
toward the middle of two extremes.
Building applications is an option that never really stopped.
Nearly two-thirds of all software capital spending is used either
to customise existing applications or to build new ones. Open
source, offshore development and web services are bringing build
back as a mainstream option.
The main consideration with open source is how fast it will take
off. Smart users are looking at how far and fast they can take open
source. Smart buyers are looking at how to use it to their benefit
(as well as to the detriment of competition). But it is early days
and by no means a simple option. Google is a good example of a
smart buyer. It claims to have the world's largest running Linux
cluster, with more than 10,000 servers.
Meanwhile, Europe is not leaping to adopt Linux as its primary
operating system, although a recent AMR Research survey revealed
some pockets of interest. Microsoft dominates the European OS
environment, but Linux is making inroads, particularly in Germany.
The Microsoft Windows environment clearly dominates the SME
segment. Across Europe, almost 75% of users sampled indicated that
their primary OS was Windows, followed by the IBM mainframe
environment (10%) and Linux (9%).
Germany is the test-bed for Linux in Europe and the adoption rates
there are among the highest observed in AMR's global study - even
higher than Brazil (11%). Possible explanations include SAP's
success with its packaged product offering for the SME market,
which incorporates SAP functionality on the open source MAXdb
database, and the presence of Linux provider SuSE in Germany may
account for more widespread adoption domesticly.
The next leg of the trend towards build is offshore development.
One of the main rationales for not wanting to build applications is
the expense and amount of time it takes. The ability to
significantly cut the labour cost of coding begins to remove one of
the largest impediments. But it does not remove the costs and
burden of long-term support.
Development labour costs can be more than halved by moving
development offshore. Internal IBM figures leaked to the Wall
Street Journal in January showed that the internal billing rate for
a US programmer with three to five years' experience was $56 (£30)
an hour, compared with $12.50 for an equivalent coder in
China.
ERP and supply chain supplier Infor has just 40 people in India. It
is using them to assist in its ERP consolidation activities. Not
only does it gain 2:1 savings on cost (an Indian programmer can
earn about $30,000 a year), the company has found that the quality
of the staff in India is very high.
When offshore work is combined with open source, as offshore
provider Aztec Software is doing, the opportunity to cut costs and
rapidly deploy applications increases dramatically. Offshore firms
are also pushing into the application space with frameworks in a
given vertical market that are about 50% complete and then
customised. Think Andersen's Mac-Pac, but with a fixed price
contract.
At the same time, these companies are also looking to support these
applications - for a price. With most suppliers now utilising
offshore development resources, India-based developers are building
knowledge of most solutions sold on the market today. It is
estimated that Oracle, IBM and Sun have a combined total of 25,000
developers in India. Wipro, an India-headquartered software and
services company, is estimated to employ another 30,000.
Finally, if these factors were not enough, the introduction of new
tools and extension of old, using web services technology, promises
to make customisation even easier: 4GL Tools such as Progress and
Lansa are still used extensively by a large number of independent
software suppliers and service providers to enhance and extend
their solutions.
In addition, an array of new tools has been introduced that is
speeding development times. They range from delivering
web-service-based integration capabilities to composite application
building tools that will assemble applications from both customised
and widely-available components. They face some significant
challenges before the reality matches their hype, but fast and
easy-to-deploy applications are closer than ever.
At the very least web services will make application creation and
integration between business processes easier and faster than
before, though widespread adoption is unlikely to be as soon as
many suppliers claim.
Think about what you could do if you combined an open source
technology base with inexpensive, high-quality offshore labour that
is using highly productive tools. IBM recently said it will be
using an Indian developer to build J2EE banking applications and
that it will be building custom, industry-oriented software for
industry-specific middleware.
This does not mean that you will throw out your SAP, PeopleSoft or
Oracle applications to build your own ERP system, but it does mean
that there are some new alternatives for buyers as they look to
craft technology products for their business - and it does
challenge suppliers to increase their own development, or at least
to partner well.
Nigel Montgomery is director, European research, at
AMRResearch
AMR's recommendations
- Building applications is a real option. Application packages
can deliver "overkill" for the functionality required.
- Revisit your application decision tree with an option to build
applications. It was never a no-brainer to buy rather than build
applications. It is even less so now.
- Revisit old assumptions about the economies of building
applications. Do not forget to include total cost of ownership
analysis as well as support costs.
- Investigate open source possibilities. While open source
applications are in their infancy, the surrounding infrastructure
and technology are coming up the maturity curve quickly.
- Look for opportunities to build uncommon applications around
this technology base as well as to use open source applications for
tactical, non-critical business processes.