Application security flaws introduced early in the design life cycle are giving rise to easily exploitable defects which can easily be prevented.

That's the main conclusion of an evaluation of 45 e-business applications by security consultancy @stake. It says the current state of application security is "grim".

It found that nearly half of application security defects - 47 percent - are both readily exploitable and could cause significant loss of reputation or customer revenue, but the defects were entirely preventable. The consultancy found that the best-designed e-business applications have 25 per cent more security defects as the worst.

A selection of wireless application, off-the-shelf aps and Web-transactional aps supplied by @stake's clients were put through their paces during its security evaluation.

The methodology involved a code review; a look at the architecture and design of applications; and attack simulations. As security testing was carried out under non-disclosure agreements it is unclear when issues will be highlighted to the wider community.

@stake's testing reveals some worrying trends in application insecurity.

These include: insufficient rigour in checking user input, a problem that can lead to all too common buffer overflow attacks; and a lack of secure authentication and access control features within applications. User session security proved to be the "Achilles heel" of many of the apps analysed.

Nine classes of common security flaws can make applications insecure, according to @stake's research. These are: administrative interfaces authentication/access control, configuration management, cryptographic algorithms, information gathering, input validation, parameter manipulation, sensitive data handling and session management.

Avi Corfas, vice president for @stake in EMEA, said the problem are more to do with the way applications were designed rather than the platform they run on. Somewhat controversially, he claimed 70 per cent of the problems he identified were in design and only 30 percent were in implementation.

Anecdotal evidence from other security consultants suggests the vast amount of security risks is caused by incorrect implementation, for example in interaction between applications and how they are installed. That's to say nothing of failure to apply patches, of course, which @stake points out would be far less of a problem if developers addressed security problems early in the design process.

"Many companies treat security as 'penetrate and patch' rather than employing secure software engineering practices that would have produced a safer application from the start," said Andrew Jaquith, program director, @stake.

In order to benchmark application security best practices, @stake compares and contrasts the top and bottom performers in its study as measured by business risk.

The six areas that differentiate the best from the rest are: early design focus on user authentication and authorisation; mistrust of user input; end-to-end session encryption; safe data handling; elimination of administrator backdoors, mis-configurations and default settings and security quality assurance.