If you were in charge of a large team implementing substantial software systems how would you ensure that the product was both reliable and finished on time?

Answer 1

It's been said: you can have your software good (quality), fast (on schedule), and cheap (within a limited budget) - but you can only pick two. To have the software be reliable and on-time you must make sure your plan includes sufficient resources - both people (with the right skills) and the option to pay overtime when necessary to relieve bottlenecks. The project plan should have built-in time and resource reserves to apply when Murphy's law comes into play so that bottlenecks don't force a choice between meeting schedule and producing a reliable product.

In project planning you have to identify at the outset what the dependencies are: all your finish-to-finish, finish-to-start, start-to-start, and start-to-finish dependencies. This makes it possible to identify key milestones early and plan for reviews at those points to keep the project on schedule. One important task to include is proper intermediate testing. You cannot consider a sub-task done until the code has been tested - but the testing cannot start until the coding is at least partially done. Defects found during testing must be re-worked and regression testing must be done once the defects are corrected.

If this is starting to sound a bit involved then you can see the value of having well defined change management and defect tracking/resolution processes. With these processes operating correctly, the project manager and the large team they are directing should always have a good handle on whether the product is on-schedule and whether it is meeting its quality (reliability) targets. If the schedule slips or reliability issues crop up, the reserves that were included in the plan at the beginning are available to keep things moving. Note that if the reserves were NOT included in the initial plan then the ad hoc "crap! what do we do now?" method of management may also result in unplanned overall schedule slip, degraded quality, AND increased costs.

==========

For reliable, you can obviously test it by making your own fake Trojan.

for finished on time, get an estimated date, then add 1 or 2 weeks just in case of any delay and just make sure to finish it a few days before release.

NOTE:

Obviously, no anti-virus can be 100% reliable as proffessional hackers can always find a loophole, just like when psn got hacked