YOUR OWN WORST ENEMY IS YOU…

From NetworkComputing Magazine Vol 16, Issue 6 - November/December 2007

Bill Beverley of F5 Networks suggests that breaking down silos can lead to more effective deployment of applications

Like people, even though many applications appear the same on the outside (delivered via HTTP in many cases), they're very different on the inside, and may need to be treated differently. But most networks are merely application aware. They know some amount of detail about, say, Microsoft SharePoint, but they aren't intimately familiar with all the little nuances of that application, and how it would best be delivered.

But more should be known. Network teams should spend time with each application and understand all of the intimate details about how best to deliver it. If network engineers can spend time up front learning these nuances and peculiar quirks, and develop policies before deployment, this will help ensure better results without interfering with the actual delivery process.

This last point is the nub
The biggest inhibitor to successful application deployment has not been the technology per se, but the people and processes that develop and deliver the whole project. Developing and rolling out an application over the web on a large scale is not just about what features it should have. It involves many different parts of large organisations interacting with each other - or at least it should.

Problems stem from developers not consulting with security teams, with the Architecture and Operations teams not co-operating.

These issues have serious impact on successful application delivery. A recent survey, conducted by Forrester in North America (published in August 2007 ) , found that a quarter of application deployments in the previous twelve months suffered a significant problem, mostly in terms of performance, but also comprising project overrun and budget overspend. A huge 71% of all deployment issues occurred in production environments; for which read, business impact.

What conclusions should be drawn from this?
Can the functional areas within organisations work better to deploy better? The Forrester survey suggests that people want to improve - 72 % of respondents claim they would accommodate organisational change in order to adopt technology that would improve its deployment quality.

Forrester go on to highlight an emerging solution, which they term application delivery infrastructure, defined as, "Technologies that streamline the connection of any user to any application, by minimising deployment burdens, reducing management costs, optimising performance, and increasing security."

This has merit in many ways
A set of tactical technologies that help overcome planning and execution deficiencies, with the end goal of making application deployments work better, is going to be attractive to project leaders. Application re-engineering is massively expensive. Infrastructure costs big money. Both take time to do right. If there's something that is a less expensive, less time-consuming way of making the application work, then it simply has to be considered.

A better approach
What if all the silos whose lack of communication in the first place created the problems, got together at the beginning of the project? This is starting to happen. People with job titles that intimate responsibility, across the gamut of IT, for effective development and deployment of applications, are starting to crop up, especially in the big banks.

They are fostering a way of working that involves collaboration between all affected parties, and they are using technology that is squarely aimed at improving application deployments across networks, right from the beginning of the project. Actions like this point the way to how application delivery should be approached, because this results in a vastly improved user experience. What is the point of an application that is insecure, that is unusable by its audience; that is slow, unwieldy, or crashes every five minutes? Hardly business aligned or even serving is it? NC