Lecture Software process improvement: Lesson 30A provide students with knowledge about: software process benchmarks; macroeconomic benchmarks; economic benchmarks; corporate IT benchmarks; customer satisfaction benchmarks; project-level benchmarks;... Please refer to the detailed content of the lecture!
Trang 1Lecture # 30A
Trang 3• Benchmarks compare a company against
industry norms
Trang 4• All of these factors are related to software process improvement initiatives
Trang 5• Benchmark studies often include
supplemental information on specific tool suites, programming languages, and formal methods utilized
• Thus, benchmark studies and assessment
studies overlap to a degree
Trang 6• Benchmarks are primarily comparisons
between a specific company and other
companies in the same kind of business
Trang 7• Benchmark studies or comparisons of
quantitative data are older than the software industry. Indeed, comparisons between the companies have been taking place for
centuries. E.g., studies of market shares,
staff compensation levels, executive pay,
and customer satisfaction have long been
performed
Trang 8• Software benchmark is a formal comparison
of software methods and results against
those of other organizations
Trang 9• Software benchmarks can be traced back to the 1960s, when studies of data center
performance and downtimes started to be performed
• These were soon followed by corporate
level studies of investments in information technology, compensation studies of
software personnel, and studies of software schedules, cost, and process
Trang 10• Within the software industry, there are a
number of levels of benchmark studies that have been noted to occur
• It would be a good idea to consider the
various kinds of benchmarks
Trang 11• Macroeconomic benchmarks are concerned with very largescale issues such as impact
Trang 12• Economic benchmarks are concerned with topics such as whether investments in
computers, software, factory automation, or process improvement benefit the
profitability and market shares of
companies that spend more than others
• Productivity paradox
Trang 13• Economic benchmarks are performed by
economists, universities, and by
management consulting groups
Trang 14• Corporate information technology
benchmarks deal with a host of interesting comparisons among companies
• Some of the comparisons that are studied under corporate benchmarks include
– Percent of corporate employees in IT
– Number of user supported per staff member in
Trang 16software compensation and salary levels
Trang 17Benchmarks
• Customer satisfaction benchmarks also
predate the computer era and can be traced back more than 100 years. In the context of software applications, customer satisfaction surveys for specific products are often
carried out by the vendors themselves
• However, comparative benchmarks
between classes of similar products require more extensive data
Trang 18• Projectlevel benchmarks for software have been carried out since the 1970s
Trang 19• With hundreds of companies performing
scores of benchmarks it is not surprising
that there are many variations in how the
work is performed
Trang 20• Benchmark studies can be carried out in a variety of ways, including
Trang 21• Benchmarks by means of mailed/emailed survey instruments
• Benchmarks by means of telephone surveys
• Benchmarks by means of onsite interviews
Trang 23• A good benchmark should also uncover
information on why the results are the way they are
• Also, because benchmarks are precursors to process improvement programs, the
benchmark should indicate some of the
steps that might be needed to improve the client’s standing against data
Trang 24Basic Sequence of Software
Benchmarks
Trang 25Benchmarks
Trang 26• Benchmarks are even more effective than assessments in causing companies to start software process improvement programs
• If a company is well below average ranges
in productivity and quality, this situation
presents tangible data that executives can understand easily
Trang 27• For example, the virtues of moving up from
an SEI CMMI level 1 to a level 3 might be abstract and difficult to explain, but the
need to improve productivity rates from
five function points per staff month to ten to reach industry norms is fairly unambiguous
Trang 28• Accurate benchmarking of software
projects was difficult for many years due to problems with the LOC software metric,
which was unreliable for studies involving multiple programming languages
• The advent of the function point metric has opened a door to more accurate benchmark
Trang 29• Benchmark studies can explore a number of different topics
• Very interesting software benchmarks can
be carried out at the project level. These
projectlevel benchmarks can generate very useful information. However, projectlevel benchmarks require a great deal of care to ensure an applestoapples comparisons
Trang 30• What are bestinclass productivity rates in our industry?
• What are bestinclass development costs per function point in our industry?
• What are bestinclass development
schedules?
Trang 31• What are bestinclass quality levels in our industry?
• What are the best development processes
for software like ours?
• What are the best development tools for
software like ours?
Trang 32• What does it cost to move from SEI CMMI level 1 to SEI CMMI level 3?
• What are the differences in results between SEI CMMI level 1 and CMMI level 3?
Trang 33• As can be seen from these very common
questions, the most interesting topics from benchmarks are those associated with best practices and bestinclass results
• Averages are interesting and useful, but
benchmark studies are often commissioned
by companies with goals that are to be
much better than average
Trang 34• For benchmarks to be useful, or even for the comparisons to be valid, the projects that are compared during the benchmark have to be
similar in size and in nature
• For benchmarking purposes, it is not a normal practice to compare unlike software
applications, such as comparing the
productivity level of a small Web applet with
Trang 35Benchmarking Overview
Trang 36Personnel Processes Technology Environment
Productivity Rates Quality Levels WhereYou Are Department ModelsProject Profiles WhyYou Are MEASURING
ASSESSING
Trang 37Summary
Trang 38• Software Assessments, Benchmarks, and
Best Practices by Capers Jones (Chapter 3)