Community Collaboration for ERP Implementation ... - IEEE Xplore

4 downloads 2243 Views 1MB Size Report
case-based knowledge repository system that sup- ports ERP package implementations; prototype deployment at an ERP software community Web site shows ...
focus

cooperative and human aspects of SE

Community Collaboration for ERP Implementation Harris Wu and Lan Cao, Old Dominion University

A Web 2.0 knowledge repository system can reduce costs, improve quality, and lower the risks of enterprise resource planning implementations.

E

nterprise resource planning (ERP) implementation solves business problems by customizing and integrating off-the-shelf enterprise software packages. Effective, efficient implementations can bring a customer millions of dollars in returns, improve implementation consultancies’ profit margins, and sustain software vendors’ growth. Successful ERP implementations avoid reinventing the wheel and thus lower custom software development efforts. ERP package implementations differ from traditional software development. They rely heavily on human aspects. Experience and knowledge are critical to an implementation’s success. In fact, when facing a problem situation, implementation professionals frequently practice case-based reasoning: 1. explore the past cases (experiences), 2. find a similar case, and 3. reuse the solution for the past case in the new problem situation.1 When the same software package is implemented at many sites, case-based reasoning can significantly improve quality, reduce costs, and lower design risks. However, a case-based approach can be only as good as the cases’ knowledge base. The main challenge of utilizing the case-based approach is the lack of support for building and growing the knowledge base. We created Epics (ERP Package Implementation Community Support), a Web 2.0

48

IEEE SOFT WARE

Published by the IEEE Computer Society

case-based knowledge repository system that supports ERP package implementations; prototype deployment at an ERP software community Web site shows promise. We believe that utilizing the collective power of a large group of people can help accomplish higher-quality ERP implementations at reduced costs with lower risks.

The Promise of Web 2.0

A new trend is to utilize the “free brains and labor out there” among Internet users by focusing on collaboration and cooperation among individuals in communities across organizational boundaries. Harnessing social intelligence—utilizing individual users’ knowledge and efforts—is a core competency of Web 2.0.2 We see such Web 2.0 phenomena everywhere: collaborative spam filtering, Amazon.com’s book recommendations, Wikipedia.com (the world’s largest encyclopedia), and even CNN. com’s attempt to create user-contributed news. ERP software vendors also have tried to build online communities to reduce support and imple0 74 0 -74 5 9 / 0 9 / $ 2 6 . 0 0 © 2 0 0 9 I E E E

mentation costs. For example, the two largest ERP vendors, SAP (Systems Applications and Products) and Oracle, have built the SAP Developer Network (www.sdn.sap.com) and the Oracle Technology Network (www.oracle.com/technology/ community) respectively. Potentially, a large pool of social intelligence remains for ERP implementations to collect and utilize (see the sidebar “ERP Market Size”). Traditional collaborative software design tools focus on working together on the same project with a common goal. Participants in an ERP implementation community, however, work on diverse projects and agendas. Even so, these individuals can share their experiences by recommending solutions to each other’s problems and drawing connections between future situations and existing solutions. So far, we haven’t seen large-scale bottom-up community efforts supporting ERP implementations. However, several emerging Web 2.0 technologies seem to promise such support. Wiki systems support collaborative documentation, requirements engineering, and knowledge contribution. Wikis let users from different organizations, both customers and consultants, collaborate on requirements and design documents. A knowledge case contributed by one user can be updated and improved by others. Whereas typical software projects archive and freeze their documents upon project completion, documents in a wiki system can evolve beyond a particular software implementation so that other teams can enhance and reuse past implementation knowledge for different projects. Social tagging can help classify and enrich knowledge cases.3 Emerging social-tagging systems such as del.icio.us let users assign and share free-formed keywords (tags) to any document on the Web. Users can browse these tags or search for documents tagged with given keywords. Social tagging has gained considerable success because it requires little effort from individual users and lets them retrieve items more efficiently. Once individuals tag documents for their own benefit, their knowledge can benefit others. Social tagging has proven especially useful for multimedia documents not amenable to search, such as photos at flickr. com and videos on youtube.com. Recommender and collaborative-filtering systems can provide both peer and systematic design recommendations (see Communications of the ACM, special issue on reputation systems, vol. 43, no. 12, 2000). A recommender system can use a hybrid set of inputs including classifications and textual descriptions of problem situations

ERP Market Size Simon Jacobson and his colleagues estimated that the worldwide market for ERP (enterprise resource planning) software licenses was US$28.8 billion in 2007 and projected it to increase to US$48 billion by 2011.1 The ERP implementation market is even larger, employing hundreds of thousands in the US alone, as costs for implementation and maintenance exceed license costs.2 References

1. S. Jacobson et al., “The ERP Market Sizing Report, 2006–2011,” AMR Research, 2007; www. sap.com/solutions/business-suite/erp/pdf/AMR_ERP_Market_Sizing_2006-2011.pdf. 2 C. Jutras, “The 2007 ERP in the Mid-Market Benchmark Report: Serving the Needs of 1.2 Million Businesses,” Aberdeen Group, 2007; www.aberdeen.com/summary/report/ benchmark/4208-RA-erp-mid-market-needs.asp.

and past design cases, as well as individual design recommendations. “Web 2.0” denotes both business models and design patterns.2 Similarly, harnessing Web 2.0 in ERP implementations involves new business models and supporting technologies.

Requirements from ERP Implementation Communities

Large software vendors and implementation consultancies have many implementation projects and could therefore benefit from synergistic efforts. For example, 10 large Oracle ERP consultancies had hundreds of implementation projects annually.4 To understand how Web 2.0 could benefit ERP implementations, we conducted a case study of 16 companies involved in ERP implementation. These companies included software vendors, consultancies, and customers covering the whole ERP ecosystem.5 All the companies had existing internal knowledge management tools such as document repositories. Although most of the companies were interested in Web 2.0, few had Web 2.0 initiatives. Our objective was to understand the characteristics and challenges of ERP implementation to guide the development of a Web 2.0 knowledge repository system. Our study identified several traits unique to ERP implementation, which we discuss in the paragraphs that follow. Another empirical study of ERP customizations by Yvonne Dittrich and her colleagues confirmed these traits.5 For each trait, we identified opportunities, challenges, and what features a knowledge repository would need to address them.

Common Solutions for Different Contexts Consultants frequently seek to reuse solutions in new contexts. The challenge is to retrieve relevant solutions. For example, one company’s repository November/December 2009 I E E E S O F T W A R E 

49

A key challenge is finding an existing solution, which is half the success.

contained many documents scanned in TIFF format, which isn’t easily searchable or reusable. For developers to reuse past solutions, a repository must capture and store documents in editable formats. The repository needs to contain proper metadata to make documents searchable, especially nontextual documents. The repository should share the documents by default, which is critical for many Web 2.0 tools.2 Moreover, everyone in an implementation team should be able to contribute to the repository. Access control lists must maintain proper data security and privacy mechanisms. Code names and wildcards must mask or replace customer identities and other proprietary information. To facilitate reuse, the repository should let users easily assemble documents and incorporate them into new projects.

Individual Consultants’ Knowledge and Experience

Gap-Filling Rather Than Building from Scratch

The repository tool needs to provide systematic or third-party solution recommendations for requirements situations to avoid personal bias, give a second opinion, and present the solution space.

Instead of the traditional paradigm that involves 1. requirements gathering, 2. design, and 3. implementation, ERP implementation typically involves 1. matching a requirement to an existing solution, 2. gap analysis, 3. reengineering the business process to reduce the gap, and 4. modifying the existing solution to fill the gap. A key challenge is finding an existing solution, which is half the success. Most customizations must follow the ERP software vendor’s predefined customization guidelines.5 Only on rare occasions and after extensive review and customer approval are implementation teams willing to perform core modifications that modify the ERP software and typically break the software warranty. Seemingly simple requirements or designs can prove costprohibitive or technologically infeasible. Both customers and implementation consultants take on significant risks when trying to achieve novel functional goals without prior success cases. Projects often revise or compromise original requirements owing to technology constraints. To minimize risks in ERP implementations, the repository must support effective knowledge exploration. Users must be able to explore and search knowledge cases from a variety of perspectives. 50

IEEE SOFT WARE

w w w. c o m p u t e r. o rg /s o f t w a re

Consultants are essential in bringing in technical knowledge as well as the best practices from past or concurrent implementations. Experienced consultants are instrumental to successful implementations. On the other hand, consultants tend to suggest suboptimal solutions when they don’t have the personal experience. As an IT director of an ERP customer commented, Consultants’ first priority is not to have the optimum design; instead, they want to finish the work. On the other hand, we [the customer] want to find the best solution. But we know only the solutions that consultants tell us.

Intensive Collaboration among Multiple Organizations A successful ERP implementation involves extensive collaboration and communication among the customer, implementation consultancy, and software vendor. Collaboration allows implementation personnel from different organizations to utilize each other’s experiences. Communication history such as email archives and forum discussions contain rich knowledge of best practices and design solutions. However, collaboration is often hindered by compatibility, connectivity, and security issues across multiple organizations with different IT infrastructures and productivity tools. Communication is costly and often considered as “overhead” to be minimized. To support collaboration among multiple stakeholders, the repository tool should provide a common platform for interorganizational collaboration and communication. Moreover, the tool should let users explore and reuse knowledge embedded in electronic communications.

Epics

Through the case study, we captured key desired features for a Web 2.0 knowledge repository system to support ERP implementations. We designed Epics to fulfill these requirements and developed our prototype. Software vendors can use Epics to support a community including their employees, implemen-

Interactive components

Back-end components

Storage, editing, and assembling Search and exploration Classification and tagging

Automated case classification

Repository: documents, metadata, facet schemas

Feedback/reputation management Solution

User feedback and recommendation Change management, access control and security

System recommendation

Figure 1. Major functional components of the Epics (ERP Package Implementation Community Support) system. They include browse-based interactive components data repository storing all documents and metadata, and back-end components with automated classification and recommendation

tation partners, and customers. ERP consulting firms can use Epics as an extranet tool to support collaboration with their customers. ERP users can use the tool to support their implementation process and maintenance. Through access control and security features, Epics encourages participants to share their implementation knowledge within and beyond their organizational boundaries.

The Basic System Figure 1 shows Epics’ major components. Data repository. Organizations’ existing document repositories, issue-tracking databases, online discussion forums, and email lists have already captured much implementation knowledge. Epics can extract and incorporate these existing cases and use projects to preserve the source knowledge bases’ document structure. For example, a project would store a discussion thread and all associated attachments. Many repositories store various documents’ properties or metadata. Whenever possible, Epics would extract and store these metadata. Users can use Epics concurrently with existing knowledge bases. Interactive components. Several front-end user interface components provide knowledge reuse, exploration, recommendation, and collaboration. For example, the storage, editing, and assembling component lets users upload, download, edit, collaborate on, and assemble documents. The classification and tagging component lets users explore and enrich the repository. The user feedback and recommendation component lets users evolve the system, filter out low-quality content to reduce information

Back office Mantis (Issue tracking) phpBB (discussion forum) MoinMoin (wiki)

Search

Emails DocBook:Colla (XML conversion/assembly)

Joomla and MySQL

Translation

Joomenco (Classification, tagging and documentation)

Figure 2. The Epics prototype. Joomla is integrated with PhpBB, Mantis, and Wiki, in additional to custom components developed for Epics.

overload, and provide recommendations to peers. The interactive components are closely integrated with back-end components and the database. Back-end components. These server-side processes support front-end operations. For instance, statistical-learning algorithms classify documents into predefined categories. Feedback management collects feedback on documents and provides it as input to the recommendation process.

The Prototype Our prototype (see Figure 2) customizes and integrates several open source software packages (see the sidebar “Major Open Source Software Packages Utilized by Epics”). Modules in purple are custom-developed or heavily customized. Most custom development uses PHP as the programming language. The prototype uses Joomla as November/December 2009 I E E E S O F T W A R E 

51

Major Open Source Software Packages Utilized by Epics Joomla (www.joomla.org) won the best content management system award from PacktPub.com in 2007. Joomla is an open source solution for building Web sites and tracking contents such as text, photos, music, video, and documents on Web sites. Joomla has a large user community and a variety of free add-on modules. MoinMoin (www.moinmo.in) is an advanced wiki engine with a large user community. The software is free; many open source projects such as Apache, Ubuntu, Debian, and OpenOffice use it to create collaborative Web sites and power community Web sites. phpBB (www.phpbb.com) is the most widely used open source forum solution. MantisBT (www.mantisbt.org) is a popular application for keeping track of reported software bugs or other issues. Software development companies use MantisBT extensively.

the integration platform and main user interface. We implemented the knowledge repository in a MySQL database that stores most data except file attachments and images. Joomla manages user authentication as well as the main database connection. The prototype integrates several open source packages with Joomla and makes them accessible through Joomla menu items. The prototype interfaces with back-office file servers and email systems. Joomla also lets users contribute and search implementation-related documents. MoinMoin (a wiki) lets users create and edit documents, including diagrams and images, through the browser and thus avoids problems with incompatibility. DocBook:Collab (a Joomla add-on module) supports the concepts of “projects” and “tables of contents,” where users can dynamically assemble project deliverables from multiple documents. DocBook:Collab also supports converting between Joomla content and XML, which facilitates information sharing and interfacing with other systems. phpBB, a discussion forum tool, lets users ask questions and provide recommendations for others’ problems and situations. MantisTB, an issuetracking tool, lets users report incidents, receive support, and collaborate on resolving issues. Both phpBB and Mantis interface with back-office email systems to notify users via email. Email communications can also be automatically posted to these systems and captured by the knowledge repository to support knowledge reuse. Joomenco, a Joomla module custom-developed by our research team, supports exploration and system recommendation. Joomenco contains advanced tagging, classification, search, and recom52

IEEE SOFT WARE

w w w. c o m p u t e r. o rg /s o f t w a re

mendation functionality and extends the socialtagging concept to faceted classification, which more effectively supports exhaustive, systematic knowledge repository exploration.3 Faceted classification supports multidimensional design exploration by assigning multiple classifications to a design document along diverse perspectives such as module, technology, and geography. It lets users browse a collection of knowledge cases through category name recognition instead of the recall of keyword search, in which recall is far more difficult than recognition.6 Joomenco also includes server-side programs that automatically classify design documents on the basis of their contents and metadata. Joomenco’s interface lets users approve, reject, or modify the system-made classifications. Upon user request, the system recommends the most pertinent design cases using the content and metadata (tags, classifications, and so on) of both problem situations and past design cases, as well as past recommendations and user feedback. Instead of presenting the recommendation in exact matches or ranked lists, as other tools do,7 the system presents a clickable map with icons for problems and recommendations spaced according to their relevance and similarities. Such intuitive representations are important to solving complex problems.8 Existing features or add-on modules from Joomla and other open source tools accomplish many of the supporting functions. Feedback and reputation management are important to all online knowledge communities. Change management is important to collaboration and reuse. Access control and security are important to deploying the system in an industry setting. Translation lets users in a global community share knowledge and collaborate. Several open source software packages integrated in Epics are best of breed in their respective categories. This integrative, bottom-up approach lets Epics exploit continuous advancements of underlying open source tools. It also lets companies incorporate Epics into their existing systems. Many software vendors and consultancies have already invested in systems that support design, coding, change management, and other software engineering tasks. They can choose to adopt part of Epics, instead of the whole system, to power their online communities. In other words, Epics can complement, not necessarily replace, existing knowledge management systems and CASE (computer-aided software engineering) tools. Epics differs from a collaboration-enabled IDE targeted mainly for a single enterprise, or hosted software

collaboration portals such as Sourceforge.net and Github.com that provide setup-free, one-stop service to software communities.

The Deployment

We deployed the prototype at xTuple, an ERP software vendor. The deployment was a joint, phased effort by xTuple and our research team. xTuple has two commercial editions of fullfeatured ERP software: xTuple Standard Edition and xTuple Manufacturing Edition. The company also has an open source edition of its ERP software. xTuple has more than 100 customers ranging from small businesses to Fortune 500 companies and more than a dozen implementation consultancy partners. In 2007, it received a five-star rating from ChannelWeb9 and became a finalist in the eWEEK Excellence awards.10 In the past, xTuple communicated with its implementation partners and customers mainly through emails and discussion forum threads, hundreds of them a day. xTuple support personnel spent hours each day searching through email archives and forum threads, trying to find past solutions to a problem. Implementation partners often expended considerable effort replicating customizations already done by others. Slow implementations hampered the market adoption of xTuple.

xTuple.org In fall 2007, xTuple implemented our prototype at xTuple.org, aiming to utilize social, community power to support ERP implementations and grow the market. To maintain Web site stability and carefully address data migration and user orientation issues, the deployment has been incremental. By the end of 2008, about 80 percent of the prototype had been deployed. Menu items of xTuple.org include browse, search, issue-tracking, discussion forum, wiki, blog, and download functionalities. The company reserves certain features, such as issue tracking and advanced search, for commercial customers or partners. Proprietary documents are accesscontrolled on the basis of user logins. For example, a user interested in developing a sales commission feature in the ERP software’s sales module could search for “sales commission” and find a previously implemented customization called “advanced sales commissions.” The user could then reuse the requirements specification and implementation solutions. The system uses code names and wild cards to hide proprietary information while allowing for knowledge sharing. We interviewed more than a dozen implemen

tation consultancies and customers with online activities on xTuple.org. All interviewees had positive experiences with xTuple.org and confirmed that the system helped address the ERP implementation challenges we identified. The community-maintained knowledge repository addresses the challenge of reusing past implementation knowledge. Most interviewees found the how-to documentation on the wiki site helpful. Several used the free code examples and addon modules contributed by the user community. Several interviewees quickly overcame implementation difficulties by participating in discussion forums. Although xTuple’s support is available in the US only during daytime hours, the online community is live 24-7. And although xTuple charges hourly commercial support fees with a tiered pricing model, most features and content in xTuple.org are free. This alone constitutes major implementation cost savings. The tool’s powerful exploration and search function helps users locate the most relevant knowledge. Interviewees reported that by searching across multiple knowledge sources such as the vendor documentation, the wiki, and the discussion forum, they identified relevant problem scenarios and solutions in the repository that would have been otherwise difficult to find. For example, one consultant worked on integrating an electronic invoicing module into the ERP system for a European client. She searched the integrated knowledge repository using various criteria such as geography (Europe), module (financial), and business process (invoicing) and found several similar problems and solutions that saved her about a week’s work. Reusing solutions vetted by the community and recommended by the system reduces overdependence on individual consultants while improving implementation quality and reducing risks. Ad hoc solutions tend to be inferior to solutions well-tested by the community. For example, an interviewee searched for automated installation and adopted an automated package installer following the system’s recommendation. The installer minimized user errors and installation failures compared to his in-house-developed installation scripts. Several interviewees searched on xTuple. org to see what wasn’t feasible and what had been successfully accomplished, which reduced risks in their implementations. A common Web-based collaboration platform and integration with emails allow collaboration and communication across multiple organizations. For example, the wiki component allows November/December 2009 I E E E S O F T W A R E 

53

Software vendors should start building online communities with in-depth collaboration support.

WYSWYG (what you see is what you get) editing, thus avoiding incompatibility among different word processors. Users don’t necessarily access the Web site through a desktop browser; emails can perform such functions as posting and replying. Several interviewees had subscribed to discussion threads through email notifications. One enjoyed using his Blackberry to fulfill most of his daily project management and issue resolution responsibilities. Epics facilitated not only ERP package implementations but also ERP software development. New software releases have incorporated many features such as requests, solutions, bug fixes from the discussion forum, and the issuetracking system. Another example of social power is that volunteers have translated xTuple ERP software (screens, menus, and help files) into eight languages since prototype deployment, using the system’s document translation feature. Higherquality, richer xTuple ERP package features, in turn, led to shorter cycles, lower costs, and higherquality implementations.

sultancies and even freelance consultants are motivated to pay partnership fees and commissions.

Results

uman aspects play a critical role in ERP implementation. As Dittrich and her colleagues point out, ERP implementation quality depends largely on how the implementation personnel’s knowledge and past experiences are reused and communicated.5 Although knowledge reuse, collaboration, and communication are important for all software projects, they’re particularly critical in ERP implementation owing to its unique challenges. Emerging Web 2.0 technologies and business models can be deployed to address these challenges and deliver high-quality implementations at reduced costs with lower risks. The lessons learned from our research apply to ERP or packaged software communities in general. Software vendors should start building online communities with in-depth collaboration support if they haven’t already done so. Implementation consultancies should explore how to exchange knowledge with vendors, peer consultants, and their clients, instead of withholding the knowledge for themselves. Software customers should explore what the community has to offer when making implementation decisions.

With low implementation costs and an attractive return on investment, xTuple has gained rapid market adoption, and the software community has grown dramatically. The xTuple open source edition has been downloaded more than 10,000 times per month for the past six months. LinuxWorld 2008 named xTuple ERP 3.0 Best Business Application. Interesting Web 2.0 business models emerged from the prototype deployment. For example, several companies needed a complex customization to support returned merchandise. They formed a consortium that collaboratively defined detailed requirements and paid xTuple to develop the feature. xTuple incorporated the new feature in its next release. This transaction model satisfied urgent customer needs with a quick turnaround and has been repeated for several other complex customizations. Without the Web 2.0 system, the customizations would have been invented separately at several customer sites, likely as core modifications, which would have broken the software warranties. The consortium model is a fine example of community power producing well-defined requirements and resulting in new ERP software package features that benefit the whole community. In addition, a significant share of xTuple revenue now comes from partnership fees and commissions. Because certain features of the Web 2.0 system are available to partners only, implementation con54

IEEE SOFT WARE

w w w. c o m p u t e r. o rg /s o f t w a re

Ongoing Work Implementing the Web 2.0 knowledge repository at xTuple.org is an ongoing process. From a technology perspective, our current focus is to move from Joomla and MySQL to Drupal and PostGres as platforms, because the latter seem more robust and enjoy better open source community support. We’re integrating xTuple’s ERP software with xTuple.org, so that users can directly create requests and explore the knowledge repository in xTuple ERP software. We’re improving the classification/tagging component to make it available for public use. We’re refining back-end features such as automatic communication classification and systematic solution recommendation. From a business perspective, we’re developing a market exchange for reusable implementation solutions and designing tiered partnership and certification models that reward repository knowledge contributions.

H

Acknowledgments

US National Science Foundation award 0725277 partly supported research in this article. Prototype implementation and deployment are joint efforts by the research team and xTuple. We appreciate xTuple’s collaboration and generous support.

About the Authors Harris Wu is an assistant professor of information technology and decision sciences at Old Dominion University. His primary

interest is social computing. Wu received his PhD from the University of Michigan at Ann Arbor. Contact him at [email protected].

Lan Cao is an assistant professor of information technology and decision sciences at Old Dominion

University. Her major research interests are agile software development and software process modeling and simulation. Cao received her PhD from Georgia State University. Contact her at [email protected].

HO W T O R E A CH U S Writers For detailed information on submitting articles, write for our Editorial Guidelines ([email protected]) or access www.computer.org/software/ author.htm.

References 1. M.L. Maher and P. Pu, eds., Issues and Applications of Case-Based Reasoning to Design, Lawrence Erlbaum, 1997. 2. T. O’Reilly, What is Web 2.0? O’Reilly, 2005. 3. S. Golder and B. Huberman, “Usage Patterns of Collaborative Tagging Systems,” J. Information Science, vol. 32, no. 2, 2006, pp. 198–208. 4. L. Herbert, The Forrester Wave: Oracle Implementation Providers, Q1 2008, tech. report, Forrester, 2008 5. Y. Dittrich, S. Vaucouleur, and S. Giff, “ERP Customization as Software Engineering: Knowledge Sharing and Cooperation,” IEEE Software, vol. 26, no. 6, 2009, pp. 41–47. 6. G.H. Bower et al., “Hierarchical Retrieval Schemes in Recall of Categorical Word Lists,” J. Verbal Learning and Verbal Behavior, vol. 8, 1969, pp. 323–343.

Letters to the Editor 7. D. Bridge et al., “Case-Based Recommender Systems,” Knowledge Eng. Rev., vol. 20, no. 3, 2005, pp. 315–320. 8. K. Hentschel, Mapping the Spectrum: Techniques of Visual Representation in Research and Teaching, Oxford Univ. Press, 2002. 9. M. Morejon, “Review: Open-Source ERP Scores,” ChannelWeb, 23 July 2007; www. crn.com/software/201002334. 10. eWeekLabs, “7th Ann. eWeek Excellence Awards: The Winners,” eWeek, 21 June 2006; www.eweek.com/c/a/IT-Infrastructure/The7th-Annual-eWEEK-Excellence-Awards-TheWinners.

For more information on this or any other computing topic, please visit our Digital Library at www.computer.org/csdl.

Send letters to Editor, IEEE Software 10662 Los Vaqueros Circle Los Alamitos, CA 90720 [email protected] Please provide an email address or daytime phone number with your letter.

On the Web Access www.computer.org/software for information about IEEE Software.

Subscribe Visit www.computer.org/subscribe.

Subscription Change of Address

Call

for Articles IEEE Software seeks practical, readable articles that will appeal to experts and nonexperts alike. The magazine aims to deliver reliable

Send change-of-address requests for magazine subscriptions to address. [email protected]. Be sure to specify IEEE Software.

Membership Change of Address Send change-of-address requests for IEEE and Computer Society membership to [email protected].

information to software developers and managers to

Missing or Damaged Copies

help them stay on top of rapid technology change.

If you are missing an issue or you received a damaged copy, contact [email protected].

Submissions must be original and no more than 5,400 words, including 200 words for each table and figure. Author guidelines: www.computer.org/software/author.htm Further details: [email protected]

www.computer.org/software

Reprints of Articles For price information or to order reprints, send email to [email protected] or fax +1 714 821 4010.

Reprint Permission To obtain permission to reprint an article, contact the Intellectual Property Rights Office at [email protected].

November/December 2009 I E E E S O F T W A R E

55