Download presentation
Presentation is loading. Please wait.
Published byJessica Reynolds Modified over 9 years ago
1
Critical Chain A Novel by Eliyahu Goldratt Copyright 1997
2
Product lifetimes are diminishing fast u Projects to create new products must be shortened drastically u Consider Modems, cellular telephones, VCR’s, PC’s,
3
Frequently projects go awry with… u Budget overruns u Time (schedule) overruns u Compromising functionality or content
4
A Scenario u You have been asked to do two tasks: TASK A ---> TASK B, in sequence u TASK A whose average duration is 10 days u TASK B whose average duration is 10 days u Estimate how long it will take you to complete these two tasks in sequence? u What is the probability that you will finish these two tasks in 20 days?
5
Everybody overestimates the time required to do their task u According to Goldratt u This is called SAFETY u Does any body (in the book) want to talk about how much safety they put into their estimates? NO u Is this true in software development? NOT ALWAYS u It is if you have an expert doing the estimating, who really knows how long it will take him
6
Parkinson’s Law u The time required to complete a task fills up the time allotted for the task. u Why?? Because of our business cultures u If you finish early, you are given more work u If you finish early, your future work estimated will be cut u Our business cultures dis-reward early finishes u Rakos: expert estimates are amazingly accurate
7
What happens after that--a possible scenario u The team leader adds safety time to the task to cover his responsibilities u The project leader adds more safety time u The project manager may add still more safety time
8
The project manager must stay focused u Or the project will not be finished on time, within budget u This means applying the Pareto principle u 80% of the benefit comes from 20% of the activities u By the time progress reports indicate something is wrong, its usually too late
9
What about progress reports? u Progress reports tell you that 90% of the project is finished in 90% of the required time. u However, another equal period of time is required to complete the remaining 10%
10
It is hard to stay focused when: u There are too many project paths on-going, in parallel u There are many critical or near critical paths u There are many projects ongoing
11
Measurements are a major problem with projects u Measurements should induce the parts to do what is good for the system as a whole u Measurements should direct managers to the point that needs their attention u So often it occurs that we measure the wrong thing. u The wrong measure leads to wrong behavior u Tell me how you measure me and I will show you how I behave
12
More Measurements
13
Projects are like chains u Each task in sequence is a link in a chain u Each link has two things u weight, to which cost is analogous u strength, to which throughput is analogous
14
Cost vs Throughput u Goldratt maintains that management in the cost world is a mirage u efficiency becomes paramount u local improvements are necessary to get global ones u Goldratt suggests the managers should manage in the throughput world, a totally different paradigm u must find the constraint--the weakest link u concentrate on that
15
Remember the five steps of TOC u IDENTIFY the project constraint--the critical path u Decide how to EXPLOIT that constraint u SUBORDINATE everything to that decision u ELEVATE the systems’ constraint to every thing else u GO BACK to step 1, and find another constraint
16
Remember u IDENTIFY u Decide how to EXPLOIT u SUBORDINATE u ELEVATE u GO BACK
17
Safety u Safety is however much time is added on to a task beyond its mean time of completion
18
Probabilistic task durations u Late finishes tend to accumulate and may increase the length of the project u Early finishes do not show up u This explains why safety disappears
19
Other problems with safety u Is wasted by the “student syndrome” u Basically, this is procrastination u Is wasted by multitasking (a person who works on several tasks at the same time) u With each change of task, a set up is required u Is wasted by dependencies between steps u These dependencies cause delays to accumulate, but advances are wasted u Delays get passed on; advances don’t
20
Problems other than safety u Early start vs. late start u Existing measurements are worthless because they are based on a cost world mentality u Existing measurements do not take into consideration the critical path
21
Early Start vs. Late Start
22
How much Safety is there likely to be? u Will project professionals admit how much safety they are putting into their estimates? u What happens when these professionals are asked to cut their durations by 10%, next time? u These professionals want to be 100% sure of getting finished on time u Therefore, the durations are likely to be twice as long as they should be u So CUT THEM IN HALF
23
Problems with subcontractors u How are contractors selected?? u Lowest bidder u How should contractors be selected? u It should be noted that contractors make their money on the changes that are requested after the fixed-price contract is signed u Is lead time as important as price??? u How should contracts with subs be written?
24
Solutions u Take the safety out of the individual tasks and put it at the end of the critical path in the time buffer, called a project buffer u This means making the tasks roughly 50- 60% as long as they would otherwise be.
25
More solutions u At the point where each feeding path intersects with the critical path, place another time buffer, called a feeding buffer. The feeding buffer protects the critical path from delays occurring in the corresponding non-critical paths. u When resources are needed on the critical path, these resources are advised ahead of time exactly when they must make themselves available. When that time comes, they must drop everything else and do the required critical tasks as quickly as possible—NO STUDENT SYNDROME, NO MULTITASKING.
26
Measurement solutions u Measure progress only on the critical path; what percent of the critical path we have already completed. u Have the project leader or team leaders measure progress on non critical paths in terms of unused buffer days
27
Shrinking the task time: Effects u There is less procrastination u There is much more focus u There is less multitasking
28
What are the ramifications of a delayed software product, intended for commercial sale? u Less market share u Less profit; maybe no profit u Lower analyst profit expectations u Declining share price u Out of business? u How many firms has Microsoft driven out of business? u Ask Philippe Kahn (founder of Borland) what the implications of getting a product late to the marketplace are
29
Programmers have a tendency to… u Add bells and whistles that no one needs
30
What happened to Professor Silver IN THE BOOK Critical Chain?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.