|
Quality in Configurations |
|
|
|
|
Monday, 16 April 2007 |
Ask anyone responsible for delivering and supporting business applications
about quality requirements and most will reply with a common mantra: ‘Five
nines reliability is the expectation of the business.' But while this goal is frequently attained
for hardware servers and networks, it's another matter entirely for most
business applications. Whether you call
them ‘n-tier applications' or ‘composite applications' or ‘wiggly apps' as Gartner's
Research VP Ronni Colville calls them (my personal favorite cognomen), that acme
of quality and reliability eludes most businesses.
But it's not for lack of effort. Significant resources are applied to the
quality challenge across the enterprise.
However, the bulk of these resources often go to ensuring quality in the
front-end of the application and in various components of the application
infrastructure stack. The configuration
parameters that are used to configure the application infrastructure receive
comparatively less attention.
This is a common bias that occurs not only in IT development and support
environments but everywhere. More resources are applied to ensuring quality in
the front-end of the application because that's what all of your business
customers actually see. But often,
the things you can't see cause
enormous problems. And these are usually
time-consuming and expensive to track down.
Think about this analogy: Who
wouldn't rather spend money on dressing up a room with some new and expensive
lighting fixtures? Everyone will see
them and appreciate the impact they yield.
But what about the wiring in the walls that connects to the main service
in the basement? Without quality there,
your investment in the front-end is ill-advised.
This is certainly the case with configuration parameters or properties in
the application infrastructure. For all the effort spent on QA'ing applications,
much of this will be wasted unless the application is built upon a reliable
foundation (OK, enough home repair metaphors!) in the application stack. Consistency and quality in configuration
management for the application stack are essential.
Certainly, the issues are well-known to industry experts. According to leading research analysts at
Gartner, Forrester and Enterprise Management Associates, application
configuration errors are the leading cause of quality and downtime
problems-between 40% and 60% of occurrences.
mValent's own
research with customers and prospects syncs with these analysts,
showing that 73% of companies experience configuration-related downtime.
Fortunately, there are a few simple steps to promote quality in
configurations that will yield tremendous benefits to overall application
quality and reliability. Here are a few considerations for IT teams:
- Build-in configuration consistency and standards across
all of your environments (Dev, QA, Staging UAT, Production) and your various
applications
- Automate the delivery of infrastructure configuration
properties
- Develop a plan to audit your configuration standards in
use to measure so-called ‘configuration drift'
- Implement processes to remediate this drift and bring
configurations back in line with your standards.
Taking
these four simple steps will drive a higher quality level in the applications
themselves.
First,
configuration consistency delivers your developers that firm foundation upon
which they develop and deliver their application code. By agreeing on configuration standards for
the application infrastructure, and insuring that these standards are used in
all environments, enterprises experience substantial improvements in
application quality.
It's
always important to eliminate as many variables as possible in the quest for
higher quality. Development, QA, UAT, Staging and Production may all use the
same code and similar systems. But when
the application infrastructure itself is configured identically across these
environments, trouble-shooting and remediation are that much easier to
accomplish.
Maintaining
this configuration data in a common database format is an essential method to
promote consistency. Combine this with a ‘model-based' view of your
infrastructure and you have a reliable vehicle for insuring quality across
various stages in the application life-cycle before applications are delivered.
Second,
IT teams can improve quality by ensuring that these configuration standards not
only are well-codified in the central database but that they are delivered
reliably to all of the target systems. Just think about the scale of a mission
critical business application, like your company's CRM system or supply chain
management system. Working with large
enterprise customers, mValent encounters applications like these running on
large sets of servers. When you have to
deliver hundreds and thousands of configuration parameters to 40 or 50 systems,
it only takes a few errors to cause tremendous downstream problems.
Often,
companies use a mix of manual methods and home-grown scripts to distribute
configuration. But these are not
‘bullet-proof' processes. A systematic
method to distribute the configuration properties and then validate that this
has been done exhaustively and correctly will also deliver much improved
configuration quality.
Third,
a quick and comprehensive audit of your configuration properties that compares
what is deployed to your acceptability standards is critical for improving
quality. This is a very useful way to
avoid problems before they occur.
Again,
having codified your configuration standards and storing these in a common
database is a key first step. Then, have
tools on-hand which traverse your various systems and collect copies of the
current configuration parameters. Next,
use a sophisticated "Compare" tool to spot differences between your standards
and the current state of your world. You
will want to track the differences down to the individual configuration
parameter level, not at the file level.
These files often contain hundreds of parameters and it can only take an
error in one or two places to cause systems to perform poorly.
Fourth,
having a method to remediate the problems that you highlight through the
previous "Compare" process completes the loop for driving higher levels of
quality. It's a beautiful thing to be
able to find all of the differences in your application infrastructure
configuration settings which may be causing application problems. But it's a lot less beautiful if you cannot
act on that information in a direct and comprehensive manner. The knowledge
itself is an important precursor, but the ability to make corrections is an
essential companion to achieving quality in configurations.
For
all your best efforts to get things right when you deploy applications, we know
that IT people and developers will continue to make changes to systems during
the application life-cycle. So, tracking differences and making needed
corrections are essential quality tools.
Conclusion
With a few steps like those outlined above,
IT teams can instill quality in the configuration management process and
deliver better application reliability.
More attention paid to those things you don't see, like configuration
settings, will provide a consistent foundation for your developers and
production users.
Jim Hickey, mValent's Chief Marketing Officer, has more than 20 years of
software marketing and sales experience with emerging and growing companies. He
is a former strategy consultant with global consulting giant, Booz-Allen &
Hamilton and has an MBA degree from the Stanford
University Graduate
School of Business and a BA degree in
Economics from Harvard
College.
Trackback(0)
 |
|
Last Updated ( Thursday, 26 July 2007 )
|
TrackBack URI for this entry
Subscribe to this comment's feed
