Someone once asked me: Which the most difficult to achieve CMMI level? My answer is level 2. The answer may surprise you, I think to improve the technical and not necessarily difficult, but changing work habits is not easy. If only from the technical difficulties, there will no doubt three to four should be the most difficult. In recent years, more and more domestic software organizations through the assessment CMMI4 level or 5, although the high maturity organizations bring a lot of new ideas to improve the quality of awareness, training a group of people, but the problem still get together and children. Here is what I see high maturity implementation of the more common ten disjointed questions:
Question 1: business goals and QPPO (quality and process performance objectives) are out of line
business goals of the IT organization should be reflected in every project, and then QPPOs is a bridge between the two. When unable to articulate QPPO relations and business goals, quantify the value system has gone up. Surrogates worthless nobody really concerned.
Question 2: disconnect the intermediate target and the ultimate goal of
which is one of the important causes of quantified targets to become inoperable. When only the project (or process) ends in order to determine whether the target can be achieved, this goal must be uncontrollable. Disjointed intermediate goals and the ultimate goal of landing will lead to the baton, which is why sometimes find reasons quantitative indicators defined project start to finish unchanged.
Question 3: The selected key processes and sub-sectors of the real impact of the project out of line
probability when the data are readily available has become main concern in choosing the key sub-processes, wrong is great, SPC has become a numbers game after all. Maybe a friend heard me tell this joke:
someone saw a man child to find things in the street, and asked: "What are you looking for?"
Answer: "I come from there, too dark to see, fall !, lost wallet "
and asked:" why is not there looking for? "
a:" it's not here, there is light Well "!
question 4: QPM disjointed practice performers and designers
This is a common problem, many four five organizations quantify the system is developed by a consultant and team metrics, executives rarely intervene. Statistics and disjointed project issues, statistics may become a tool to be operated. Executives do not fully involved in the establishment of quantified system, we know not why it must be the ultimate consequences.
Question 5: QPPO, Baseline, model, SPC disconnect
remember the year when the SEI assessment audit four five, a basic requirement is the ability to complement each other among the four, to support the achievement of project objectives. When the four how these are less than a spell, I do not know that you are not crying about it or cry?
Question 6: disjointed process performance models and processes
after the year 2007, SEI to process performance models in a dominant position. Unfortunately, too many models belong superfluous, useless, controllable factor in the process of the key did not catch. SEI desired What-if scenario decision does not appear, models and process seriously out of line, causing no one to maintain the end of the assessment, use them until the next assessment before re-turn sign.
Question 7: sample data and the actual project out of touch
this is a fundamental issue common to be ignored. And the baseline model are calculated by the sample data, will have its limitations, the scope should be determined by the sample data. When the sample data is simply not similar to your project data, you will believe predictive models do? CMMI2.0MPM practice field clear emphasis on operational metrics defined, understood and used correctly it will help solve this problem.
Question 8: metrics data and the actual root cause of the problem disconnect
measurement and analysis is to find out the story behind the data, to the four later, we hope it helps guide to find the root cause of the problem. Unfortunately, this aspect of the organization is really well done rare, easily select data, ignored data show, we can not provide valuable signals and information.
Question 9: disjointed SPC, baselines, process performance models and decision-making
In the product development process, there will be a series of management and technical decisions on the structure of the high degree of maturity in terms of past experience should be some input key decisions. CMMI High Maturity is: quantitative understanding of the past, quantitative control current, quantitative forecasting future! Unfortunately, in most four, five organizations, we can not see the baseline model, SPC any shadow in decision-making.
Question 10: High Maturity management system and out of touch
if QPPO and related quantification system does not reflect the concerns of the boss, so do not expect it to really landing friends. It's worth a play that is played when the assessment. GOV hope CMMI2.0 the practice field to help owners understand the value of processes, and improve the value of his key role.
How to realize the value of high maturity CMMI is a big challenge, and I hope to be able to establish such a system: When the end of the evaluation, this stuff will continue to go down in the applicable project, as the changes will be continuously being refined. The reason is simple, it is not worth using the project in order to meet assessment requirements, but it brings, it can help us to quickly focus on core issues, in practice, constantly improve our knowledge and process, enabling organizations to enter a benign self-improvement cycle.
(Source: Old Cong lectern)
