August 26, 2011 2:59 PM
Analytics and the next frontier
Earlier this summer, I attended GigaOm's
Structure 2011 conference in San Francisco. It was a
well-attended two-day event that focused on cloud computing. Most of the big players were all there --
from VMWare's CEO Paul Maritz, Amazon's CTO Werner Vogel to Accenture's very
own Gavin Michael.
One of the sessions was a panel of
representatives from five VC firms discussing whether cloud is just another
tech bubble in the making. Towards the end, a member of the audience asked the
panel what they saw as the next wave of technologies beyond cloud computing and
big data. Not surprisingly, different panelists gave different answers but a
few common themes emerged - one of which was analytics.
Some of the panelists construed analytics as being the
way it is applied to the management of IT infrastructure, such as predicting
changes to the network or data requests before they actually happen. Others believed that applications powered by
analytics will be the biggest users of cloud computing and that the
infrastructure to support large scale analytics will become very cheap and the
use of analytics will be pervasive.
I found these answers intriguing because for the past
several years, I have come to believe that analytics will drive the next wave
of change that will transform how we develop and architect systems. As computing infrastructure and the necessary
software components for analytics are getting cheaper and more readily
available, it is ripe to be applied in the tools, frameworks and processes we
use to develop software.
Let me give two simple examples. Remember the days when
you had to run a profiler to capture execution timing, sift through the logs to
identify your application’s performance bottleneck? Remember how time consuming
the entire process was? With the combination of test automation tools to drive
the application’s front end, and using machine-learning clustering algorithms
to automatically sift through the execution timing logs, you can create an
environment that can continuously execute the application, profile it and sift
through the execution profiles to pinpoint the location of the bottleneck - all
with a few mouse clicks.
Another example: today’s integrated development
environments (IDE) such as Eclipse, Visual Studio, and Rational Jazz already
have built-in capabilities to monitor the activities within the environment.
These capabilities can be used to instrument and analyse software development
environments to objectively monitor, collect, aggregate and predict how the
software development project is progressing and whether it will hit the expected
target in terms of quality, resources and timing. What is currently missing, is
the analytical component that takes the data collected and forecasts the
direction of the project based on a certain benchmark.
Coupled with the increasing use of software development
in the cloud, I expect analytics-powered software development tools, frameworks
and environment should quickly emerge in the next several years. What do you
think? Am I too optimistic?
By Edy Liongosari, Global Director of
Research, Accenture Technology Labs
Subscribe to this blog
Email
Permalink
Print
Contact Us
