MSDN Magazine - January 2009 - (Page 41)

Notice that my calculations do not use the data for the Design and Code Logic Middle-Tier activity. Because the Logic MiddleTier activity can be performed in parallel with the Database BackEnd activity, and the Front-End activity cannot begin until both parallel activities have finished, the shorter parallel activity (Logic Middle-Tier) does not explicitly contribute to the overall time to complete the project. This type of analysis is called the critical path method (CPM) and is a standard project management technique. With the schedule mean and variance data computed, you can now compute the probability that the entire project will take longer than 30 days to complete: z = (30.00 - 28.83) / 3.50 = 0.33 p(0.33) = 0.6293 p(late) = 1.0000 - 0.6293 = 0.3707 Figure 4 Time Meta-Risk Analysis (in Days) Activity Determine Requirements Design and Code Database Back End Design and Code Logic Middle Tier Design and Code Front End Deploy Application Optimistic 2 4 Best 5 7 Pessimistic 16 12 Mean 6.33 7.33 Variance 5.44 1.78 (1) 5 3 (3) 6 8 (5) 14 13 7.17 8.00 2.25 2.78 First, you compute the so-called z-value, which is equal to the amount of time you have scheduled to finish the project (in this case, 30 days) minus the estimated time to completion (28.83 days), divided by the total task standard deviation (3.50 days). Then you take the z-value (0.33) and look up the corresponding p-value in a Standard Normal Distribution table or use the Excel NORMSDIST function (0.6293). Finally, you subtract the p-value from 1.0000 to get the probability that your project will go over schedule. You are performing a one-tail analysis here because you only care if you take longer than scheduled and don’t care if you take less time. With this example data, the probability that the Web application takes longer than 30 days to complete is 0.3707, or nearly 40%—a rather risky situation. If you think about it for a moment, this result should make sense. Your planned schedule of 26 days of development is too close to the project limit of 30 days and so you may not have enough wiggle room to account for schedule variances. Obviously, your meta-risk results are only as good as your input data—your time estimates in this case. If your input estimates are wrong, then no amount of statistical analysis can produce meaningful results. You can compute the meta-risk probability that your project will go over budget using the same technique as I’ve just demonstrated for schedule. Once you have the probability that your project will go over schedule, you can compute the risk exposure for the meta-risk if you can estimate the monetary loss due to being late. In some situations, you may be under a contract to create a software system, and the contract may have well-defined and significant late penalties. For example, suppose your contract states that there is a $10,000.00 penalty for late delivery. Your meta-risk exposure is $10,000.00 * 0.3707 = $3,707. In other cases, the cost of a late software project is too difficult to estimate beyond “very, very expensive.” But notice that even without computing a risk exposure, your time meta-risk analysis yields useful information. If you examine the data in Figure 4, you can see that Determine Requirements task has the greatest schedule variance. Therefore, simply applying additional resources early on in the project can reduce your task variance, which in turn will reduce the probability of going over schedule. msdnmagazine.com Unlike time and cost meta-risks, where the risk events can be determined in a somewhat step-by-step (although by no means easy) way by iteratively decomposing tasks into smaller subtasks, risk identification in the general case is much less mechanical. In a software development and testing environment, there are three main approaches to risk identification: taxonomy-based, scenariobased, and specification-based. A taxonomy is just a classification list. Consider the following analogy. You are going to take a trip on an airplane, so you use a standard reminder list that you use before every trip. The list con- Accelerate by up to x20: Visual Studio Builds Make-Based Builds Data Builds R&D Scripts Custom Applications IncrediBuild is an easy-to-use platform for accelerating Windowsbased processes using advanced Grid Computing technology technology. Download FREE, fully-functional 30-day trial: http://www.xoreax.com January 2009 41 http://www.xoreax.com http://www.xoreax.com http://www.msdnmagazine.com

Table of Contents Feed for the Digital Edition of MSDN Magazine - January 2009

Toolbox CLR Inside Out Basic Instincts Cutting Edge Test Run First Look� Geneva Framework Silverlight Windows Mobile Service Station Security Briefs Extreme ASP.NET Foundations .NET Matters { End Bracket }

MSDN Magazine - January 2009

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.