 |
|
|
|
 |
Newsletter |
| Subscribe to our newsletter |
|
|
|
|
|
|
|
| Latest news |
 |
|
|
| Dariel Solutions builds PC lab at Vuleka |
| Dariel Solutions provides SAPPI with a new development methodology |
Sunday Times Guest Column: eNatis crash may be due to missing safety net
Written by Lisa Lyhne
The new National Traffic Information System (eNatis) has resulted in a litany of stories about the failure of an IT project that was to drive the transport department firmly into the 21st century.
Then there is the reaction of deeply frustrated citizens, who are expected to pay for this catastrophe.
The idea behind eNatis is a good one, meeting international best practice for government service delivery. The plan is to replace an outdated system, shrink the clerical bureaucracy in favour of self-service, and to radically improve delivery at service centres.
There is no doubt the system will eventually be up and running, but it has already lost credibility with its sponsor and the confidence of the public. eNatis is not alone, however. Many large IT projects suffer similar fates, although few play out in full public gaze.
Perhaps the real issue that needs examination is why so many well- intentioned IT projects fail.
This is a list of questions the organisation and people left with the bill for a poorly executed IT project are entitled to have answered:
- Why is it so expensive?
The figure being bandied about for eNatis is R400-million. Is a fair or excessive profit being made? Were penalties included in the contract in case of non-delivery?
- Were there performance and quality specifications?
This system is a basic Crud system (Create, Read, Update and Delete). It needs to operate at high volumes, but the transactions themselves are not complex.
- What process-engineering analysis was conducted?
No system can change the behaviour and efficiency of business unless the underlying processes are understood. It may be that little attention was paid to the way in which staff handle service requests, the volume of transactions, and the nature of the information.
- What happened to incremental delivery across four years?
Modern approaches to IT project delivery advocate that pieces of the solution are delivered on a frequent basis so that they can be absorbed by the business and the system can evolve according to constant user feedback.
- Why was the pilot not extended?
There was one eNatis pilot site which performed fairly well, but the next step was a mass roll-out.
In order to mitigate the risk in a project of this scale, it would be prudent to first get it working at one centre, with a gradual roll-out, branch by branch.
It is likely that the project manager was pressured into an accelerated roll-out because the system had taken so long to develop, resulting in chaos.
- Where was the stress test?
It is incredibly difficult to predict the behaviour of large-scale IT systems.
The system may work perfectly well when testing with 20 or even 200 users, but what happens when 2000 users log on at the same time nationally? It is possible to simulate this type of situation before launch in a procedure known as stress testing. Was this done on eNatis? It seems unlikely.
- Where was the roll-back strategy?
Every implementation needs a plan, and every plan should include a roll-back strategy. When the system clearly could not deal with the load during the first week, it should have been possible to restore quickly the old Natis system to support business as usual.
- Were users at the centres trained and ready for the launch of the new system?
The staff became apologists for the system as they had to deal with an increasingly hostile public, but it is clear that they were equally baffled by the slowness of the system, and many of them appeared to be using it for the first time.
The issues raised above are not new or unique to the eNatis project, but may apply equally to any large-scale IT project.
The eNatis failure grabbed and held the headlines because it managed to fall into every trap; the biggest one was denying that there was anything wrong in the first place.
|
|
|
|
|
 |
|
|
|
