Professional Documents
Culture Documents
Why do so many challenges exist in traceability practices today? While many of these challenges can be overcome through
organizational policy and procedure changes, quality requirements traceability tool support remains an open problem. After
discussing the basics of software requirements traceability, this article shows why neither manual traceability methods nor
existing COTS traceability tools are adequate for the current needs of the software engineering industry.
Process Replication
information is usually not the one who changes occur to any elements captured in developing tiny projects? This is highly
makes use of it later. Those involved with the traceability data, the affected portions unlikely. In 1994, Orlena Gotel and
creating and maintaining traceability data of the traceability data must be updated Anthony Finkelstein [2] found that manu-
may feel that they are helping others to manually. This requires discipline and a al traceability methods were preferred in
look good while reducing their own pro- significant amount of time and effort the industry due to shortcomings in avail-
ductivity. spent on link-checking throughout the able traceability tools. It is apparent that
The easiest way to correct organiza- traceability data. Because of this, it is easy this problem still exists today because
tional problems related to traceability is for manually created traceability data to manual traceability methods are still pre-
through the use of policy and training. If become out-of-sync with the current set ferred by a significant percentage of soft-
an organization has clear traceability poli- of requirements, design, code, and test ware organizations.
cies in place and provides training on cases.
how to comply with these policies, it is Manual traceability methods are also Problems With COTS Traceability
likely that traceability will be implement- prone to errors that are not easy to catch. Tools
ed in a thorough manner consistent with Errors can arise from simple typographic Regrettably, currently existing COTS
policy [12]. mistakes, from inadvertently overlooking a traceability tools are not adequate for the
portion of the traceability data (such as an needs of the software engineering indus-
Poor Tool Support individual requirement), or from careless- try. Studies have shown that existing com-
Poor tool support is perhaps the biggest mercial traceability tools provide only sim-
challenge to the implementation of trace- plistic support for traceability [5].
ability. Even though the International
Council on Systems Engineering (INCOSE)
... the number of Surprisingly, the tools that are available do
not fully automate the entire traceability
has a survey (see [14]) that lists 31 differ- traceability links that process; instead, they require users to
ent tools claiming to provide full traceabil- manually update many aspects of the
ity support, traceability tool penetration need to be captured traceability data. This has led some
throughout the software engineering researchers to conclude that poor tool
industry is surprisingly low. Multiple stud- grows exponentially with support is the root cause for the lack of
ies have found the level of commercial traceability implementation [19].
traceability tool adoption to be around 50 the size and complexity COTS tools are typically marketed as
percent throughout industry [15, 16]. The complete requirements management
majority of the remaining companies uti- of the software system. packages, meaning that traceability is only
lize manual methods (such as manually one added feature. The traceability fea-
created traceability matrices for imple- This means that tures usually only work if the project
menting traceability), and a small percent- methodology is based around the tool
age develop their own in-house traceabili- manually capturing itself. Unless the project is developed
ty tools. from the ground up using a particular
traceability data for large tool, the tool is unable to provide much
Problems With Manual Traceability benefit without significant rework.
Methods
software projects Support for heterogeneous computing
Traceability information can be captured environments is also lacking.
manually through utilizing techniques requires an extreme Although most tools support the iden-
such as traceability matrices. A traceability tification of impacted artifacts when
matrix can be defined as a table that illus-
amount of time changes occur, they typically do not pro-
trates logical links between individual vide assistance with updating the traceabil-
functional requirements and other system
and effort. ity links or ensuring that the links and
artifacts [8]. Since traceability matrices affected artifacts are updated in a timely
are in tabular form, they are typically cre- ness by the individual capturing the data. manner [17]. This means that even when
ated using a spreadsheet or a word pro- Because traceability artifacts for large pro- tools are used, the traceability information
cessing applications table function and are jects are often hundreds or even thou- is not always maintained, nor can it always
independent of the artifacts from which sands of pages in length, such errors are be trusted to be up-to-date and accurate.
theyve captured traceability information. difficult to detect when depending on This problem is exacerbated by the fact
An example traceability matrix is shown in manual methods for error checking. that tools typically only allow primitive
Table 2. Because of these disadvantages, manu- actions to be taken (in regards to trace-
Unfortunately, manual traceability al traceability methods are not suitable for ability).
methods are not suitable for the needs of anything other than small software pro- Another issue with tools is that they
the software engineering industry. In [17], jects. Ralph R. Young stated: in my judg- often suffer problems with poor integra-
the authors found that the number of ment, an automated requirements tool is tion and inflexibility. This has led at least
traceability links that need to be captured required for any project except tiny ones one researcher to conclude that existing
grows exponentially with the size and [18]. Similarly, Balasubramaniam Ramesh traceability tools have been developed
complexity of the software system. This (in [12]) found that traceability is error- mostly for research purposes, and that
means that manually capturing traceability prone, time-consuming, and impossible to many projects are still waiting for tools
data for large software projects requires an maintain without the use of automated that do not require a particular develop-
extreme amount of time and effort. tools. Then why would nearly 50 percent ment or testing methodology [15].
Manual traceability methods are also of software companies use manual trace- Cost is another major disadvantage.
vulnerable to changes in the system. If ability methods? Is it because they are all Although the licensing fees vary per tool,
Process Replication
October 18-21
MILCOM 2009
Boston, MA
www.milcom.org
October 19-23
International Conference on Software
Process Improvement 2009
Washington, D.C.
www.icspi.com
2010
22nd Annual Systems and Software
Technology Conference
www.sstc-online.org
COMING EVENTS: Please submit coming events that
are of interest to our readers at least 90 days
before registration. E-mail announcements to:
nicole.kentta@hill.af.mil.
zational changes by groups interested in Oriented Programming, Systems, 16. Lempia, David L., and Steven P. Miller.
improving their traceability practices. Languages, and Applications. Dallas, Requirements Engineering Manage-
Poor tool support for traceability remains TX: 325-329. ment. Proc. of the National Software
an exception; this is an area that is still an 11. Cleland-Huang, Jane, Carl K. Chang, and Complex Electronic Hardware
open problem. Existing tools are costly and Yujia Ge. Supporting Event Based
Standardization Conference. Atlanta,
and provide only partial traceability sup- Traceability Through High-Level
port. This means that implementing Recognition of Change Events. Proc. 2006.
traceability is often tedious, requiring a of the 26th Annual International 17. Cleland-Huang, Jane, Carl K. Chang,
large amount of manual effort. Computer Software and Applications and Mark J. Christensen. Event-
The lack of quality tools for imple- Conference on Prolonging Software Based Traceability for Managing
menting traceability is not a technically Life: Development and Redevelop- Evolutionary Change. IEEE Trans-
insurmountable problem. The solution ment. Oxford, England, 2002: 595- actions on Software Engineering 29.9
simply involves creating cost-effective 602. (2003): 796-810.
traceability tools that improve upon the 12. Ramesh, Balasubramaniam. Factors
design and feature set of currently avail- 18. Young, Ralph R. Twelve Requirement
Influencing Requirements Traceability
able tools. Such tools would serve to Practice. Communications of the Basics for Project Success. Cross-
greatly improve the practice of traceabil- ACM 41.12 (1998): 37-44. Talk Dec. 2006.
ity in the software engineering industry. 13. Jarke, Matthias. Requirements 19. Spanoudakis, George, et al. Rule-
Tracing. Communications of the Based Generation of Requirements
References ACM 41.12 (1998): 32-36. Traceability Relations. Journal of
1. The Standish Group. The Chaos 14. INCOSE. INCOSE Requirements Systems and Software 72.2 (2004):
Report. 1994 <www.ibv.liu.se/content Management Tools Survey. 2008 105-127.
/1/c6/04/12/28/The%20CHAOS <www.paper-review.com/tools/rms/
%20Report.pdf>. 20. Naslavsky, Leila, et al. Using Scenarios
read.php>.
2. Gotel, Orlena, and Anthony Finkel- to Support Traceability. Proc. of the
15. Gills, Martins. Software Testing and
stein. An Analysis of the Require- Traceability. University of Latvia. Third International Workshop on
ments Traceability Problem. Proc. of 2005 <http://www3.acadlib.lv/grey Traceability in Emerging Forms of
the First International Conference on doc/Gilla_disertacija/MGills_ang. Software Engineering. Long Beach,
Requirements Engineering. Colorado doc>. CA, 2005: 25-30.
Springs, 1994: 94-101.
3. Dmges, Ralf, and Klaus Pohl.
Adapting Traceability Environments
About the Authors
to Project Specific Needs. Communi- Andrew Kannenberg is Hossein Saiedian,
cations of the ACM 41.12 (2008): 55-
62. a software engineer at Ph.D., is currently a pro-
4. The Standish Group. The Chaos Garmin International in fessor of software engi-
Report. 2006. Olathe, Kansas, and is neering at the University
5. Ramesh, Balasubramaniam, and currently working on his of Kansas. Saiedians pri-
Matthias Jarke. Toward Reference doctorate in computer mary area of research is
Models for Requirements Traceabili- engineering at the University of Kansas. software engineering and in particular
ty. IEEE Transactions on Software He received a bachelors degree in com- technical and managerial models for
Engineering 27.1 (2001): 58-93. puter science from the South Dakota quality software development. His past
6. Palmer, J.D. Traceability. Software
School of Mines and Technology and research has been supported by the
Requirements Engineering. Richard H.
Thayer and Merlin Dorfman, eds. New his masters degree in computer science National Science Foundation as well as
York: IEEE Computer Society Press, from the University of Kansas. regional organizations. He has more
1997. than 100 publications on a variety of
7. Heindl, Matthias, and Stefan Biffl. A Garmin International, Inc. topics in software engineering and com-
Case Study on Value-Based Require- 1200 E 151st ST puter science and is a senior member of
ments Tracing. Proc. of the 10th Olathe, KS 66062 the IEEE. Saiedian received his doctor-
European Software Engineering Phone: (913) 397-8200 ate in computing and information sci-
Conference. Lisbon, Portugal, 2005: E-mail: andy.kannenberg ences from Kansas State University.
60-69. @gmail.com
8. Wiegers, Karl. Software Requirements.
2nd ed. Redmond, WA: Microsoft The University of Kansas
Press, 2003. Electrical Engineering and
9. Boehm, Barry. Value Based Software Computer Science
Engineering. ACM SIGSOFT Soft- University of Kansas
ware Engineering Notes 28.2 (2003). 12600 Quivira RD
10. Clarke, Siobhn, et al. Subject- Overland Park, KS 66213
Oriented Design: Towards Improved Phone: (913) 897-8515
Alignment of Requirements, Design, Fax: (913) 897-8682
and Code. Proc. of the 1999 ACM E-mail: saiedian@ku.edu
SIGPLAN Conference on Object-