A CTA UNIVERSITATIS LODZIENSIS FOLIA OECONOMICA 157, 2002
Julie Cameron
UN-EDIFACT OR XML:
A CRITICAL ECOMMERCE DECISION
Q u e stio n s a re b ein g ra ise d a b o u t w h e th e r U N -E D IF A C T o r X M L m e ssa g in g sh o u ld be a d o p te d f o r eC o m m e rc e . T he c a se stu d y m e th o d is u s e d to d e sc r ib e a n d d ir e c tly c o m p a re e x p e rie n c e im p le m e n tin g m e ssa g in g in tw o in d u s try p r o je c ts o f a s im ila r size, s c o p e a n d c o m p le x ity : o n e in v o lv in g th e d e sig n a n d p ilo tin g o f U N -E D IF A C T m e ssa g e s in th e in te r n a tio n a l M a r itim e T ra n s p o rt in d u s try o v e r a o n e y e a r p e r io d 1 9 9 4 -9 5 ; th e s e c o n d in v o lv in g the d e v e lo p m e n t, testin g a n d im p le m e n ta tio n o f X M L m e ssa g e s in the S u p e r a n n u a tio n In d u s try o v e r a f o u r y e a r p e r io d 1 9 9 8 -2 0 0 2 . The c o m p a ris o n h ig h lig h ts th e im p o rta n c e a n d im p a c ts o f m e ssa g e se le c tio n a n d d e sc r ib e s th e issu e s e n c o u n te r e d a n d o u tc o m e s. The p a p e r c o n c lu d e s b y listin g c rite ria f o r u se w h e n d e te r m in in g 'h e a p p ro p r ia te m e ssa g in g p r o to c o l f o r e C o m m e rc e initia tives.
Introduction
A ustralian research undertaken for the National O ffice for the Inform ation Econom y (N O IE )1 into the outcom es o f eCom m erce projects finds failures are norm ally due to:
> Inappropriate project managem ent
> Technology: protocols, security, tools, eG atew ays > M essage Standards: uncertainty, concern
У Cost
> Lack of appropriate skills & experience
Info.T.EC Solutions Ply Ltd Sydney, Australia Telephone: 61-2-92510666 E-mail: lnfotec_solutions@ yahoo.com.au
У Slow take-up (fear o f failure, poor business case) У Failure to com plete (due participant drop out, delays).
T his reinforces the im portance for all eC om m erce initiatives o f correctly determ ining factors related to and im pacting on:
1. Project m anagem ent (people & organisations)
• • 2 3 •
2. M essage developm ent, including protocol and syntax selection (skills, technology and interoperability)
3. Im plem entation (eG atew ays, Straight Through Processing) 4. Take-up rates of trading partners (cost/benefit, ROI).
The choice o f m essaging is crucial for ail eC om m erce projects. A debate about w hether U N-EDIFACT or XM L m essaging should be adopted is underw ay. This paper provides input to this debate by com paring and contrasting the developm ent and im plem entation o f U N -ED IFA C T and XM L m essaging for industry use from a collaborative4 project m anager’s perspective. It does not address issues related to “technical superiority” but concentrates on business outcom es and consequences for the project, participants and the trading com m unity. The com parison highlights differences in effort required for each phase, the issues encountered and outcom es. The paper concentrates on the im pacts of m essaging on project m anagem ent, m essage developm ent, im plem entation and take-up rates and concludes by listing criteria for use when determ ining the appropriate m essaging protocol.
Case Studies
An abbreviated case study m ethod is used to describe and directly com pare the processes used in two industry projects o f a sim ilar size, scope and com plexity established to develop, test and im plement M essage Im plem entation
2
"Protocol” is used as a generic term the set o f rules related to UN-EDIFACT or XML messaging. The author considers that XML is currently a “language” that cannot yet be considered as a "standard”.
1 “Syntax” is the form o f the protocol (eg. XSD. XDR, DTD are examples o f XM L syntax).
4
A “collaborative” project is a project spanning a group o f independent organisations that have made a commitment (whether formally or informally) to work together to achieve mutually agreed outcomes; in which participants anticipate varying degrees o f longevity in the associations from one-off, via occasional, to consistent and long-term; and in which a moderate degree o f rationalisation or re-engineering is intrinsic. (Cameron & Clarke 1996, p l4 4 ).
G uidelines5 (M IGs). The author, as project manager, used the sam e m anagem ent fram ework and principles for both these A ustralian projects. T he projects involved:
> UN-ED IFA CT guidelines in the international M aritim e T ransport industry over a one year period 1994-95
> XM L Standards in the Superannuation Industry over a four year period 1998-2002.
I n d u s t r y a n d P a r t i c i p a n t P r o fi le
Both the M aritim e Transport (Transport) and S uperannuation (Super) Industry sectors were using paper docum entation and cheques at the tim e the projects were established. Both industries have significant im pact on the A ustralian com m unity and although governm ent has a regulatory role, neither industry is dom inated by one organisation able to enforce m essaging standards.
Transport Project
The transport industry in 1994 included over 1,000 exporters, 40 shipping lines/agents, 3 container term inal operators, 150 sea freight forw arders, over 200 road transport operators, 6 rail operators and over 30 depots and packers. The cost o f preparing and delivering each docum ent in 1994 was estim ated at $406. At least 9 types o f docum ent were required for each shipm ent. D ocum entation is transferred internationally. The retail industry that uses UN- ED IFA C T is the key interface.
An initial review of the IT capability o f project participant found that o f the 11 organisations that used packages, 5 used a product with EDI capability. One participant had used U N-EDIFACT for 5 years. T hree VAN S provided m essage exchange, translation and network services. Peer to peer EDI data exchange am ong 3 trading partners using ISDN was occurring. One participant used email to transfer data. The im plem entation o f EDI technology was understood within the industry and the use o f UN -ED IFA CT was supported internationally.
The project was I o f 3 projects that formed part o f a w ider im plem entation of EDI sponsored by Tradegate Australia, established to facilitate eC om m erce in the trade and transport industry. The scope o f the project involved the
5 "M essage Implementation G uidelines” are defined as specifications of messages, business rules, protocols and usage for use within a specified trading community.
developm ent of 7 messages. From 1 year from D ecem ber 1994 to 1995 a technical and a project m anager (supported by U N -ED IFA C T specialists) worked with the 75 A ustralian participants of the U ser G roups and 2 key users from New Zealand.
Super Project
The trading chain includes em ployers, payroll providers, SFA s, regulators (eg A ustralian Taxation Office (A TO )), banks, and hubs. S uper is com pulsory for earners in Australia. There are 213 funds, 21.7 m illion fund m em bers (an average o f 2.7 accounts for each em ployee) and 530,000 em ployers pay contributions. In June 2000 super funds totalled $477 billion with annual contributions o f $46 billion. A dm inistration costs were over $2.5 billion a year. Superannuation docum entation is confined to A ustralia. Key interfaces are the finance and banking industries, users of largely proprietary standards.
Between 1991-98 the Super Industry had developed 3 versions o f UN- ED IFA C T standards for m em ber contributions and m aintenance with the assistance o f an EDI association. Each standard was developed by Superannuation Fund A dm inistrators (SFAs) and a payroll, trialled and used for varying periods o f tim e by 2 -5 organisations before its use was discontinued due to the com plexity of translation tools, difficulties in integrating the m essaging with existing system s or lack of trading partners. F our project participants had used the UN -ED IFA CT messages before reverting to flat file for peer to peer data exchange. Hubs and payroll providers were providing aggregation services. Three organisations used web technology for data capture. All participants used email for business purposes but not for exchange o f data. No m em ber o f the industry was exchanging XM L m essaging.
The scope o f the project, sponsored jointly by three industry bodies and the A TO and NOIE, involved the developm ent o f 5 transactional m essages, workflow and paym ents conventions. From 1998 to 2001 50 volunteer organisations, formed into Special Interest Groups (SIGs) com prising about 20 business users, developed technology neutral m essages. Independently two separate groups, com prising 25 participating organisations, contributed resources to testing messages and developing M IGs from 1999 to 2001. The protocol and syntax was determ ined by these participants. T he project was staffed by a part-tim e Project D irector and a full-tim e Project M anager.
C o l l a b o r a t i v e P r o j e c t M a n a g e m e n t
From the perspective o f a Programm e M anager, the m anagem ent and im plem entation o f eCom m erce is com plex. Traditional project m anagem ent m ethods are inadequate if:
> Participant organisations are volunteers
> No single formal pow er or unified authority exists > All participants and their representative(s) must benefit > G roup cohesion must be m aintained throughout the project V All participants m ust agree with decisions
> D ecisions, schedules and specifications cannot be im posed.
eC om m erce requires the use o f a “collaborative project m anagem ent” fram ew ork and process com bined with the availability ot appropriate technical knowledge. In 1995 the author worked with Dr R oger C larke to develop a fram ew ork for collaborative project m anagem ent. T his fram ew ork is set out in “T ow ards a Theoretical Fram ew ork for C ollaborative Electronic C om m erce P rojects Involving Small and M edium -Sized E nterprises” (C am eron & Clarke 1996). Based on this research and subsequent experience both projects were managed successfully according to the fram ework and principles set out in this published paper. Additional details about project establishm ent, governance, activities and outcom es for the Transport project are set out in the EX TED I Project Report (C am eron J & Jeacle P 1995) and on w w w .tradegate.org.au; and for inform ation about the Super Project refer to w w w .superec.org.
The main im pact o f m essage selection for project m anagem ent relates to: r- Tim e taken to com plete the project (Longer elapsed tim efram es significantly
increase the risk of participant drop out, burnout and loss o f m otivation of m anagem ent and representatives as well as increasing project cost)
> C om plexity o f developm ent, testing and im plem entation
> Need for technical skills transfer if the protocol new within the industry.
M e s s a g e D e v e l o p m e n t
The m ethods used to for MIG developm ent varied in the phases used and the process adopted. A formal docum ented process described in The “SIW G M essage D evelopm ent Process M ethod 1.1” (Finkelde et al 1999), incorporating the use of data m odelling tools, was used to develop the S uper m essage to ensure rigour and consistency. This was im portant because m essages were reverse engineered from UN -ED IFA CT or were developed for additional super
transactions. Specifications were technology neutral and appropriate for UN- EDIFACT, X M L or flat file. The average elapsed tim e taken to develop one draft m essage was 5 months.
The use o f U N -ED IFA CT was determ ined by the industry prior to the establishm ent o f the T ransport project. Business analysis was used to identify the data required for the business transactions and for verifying m apping to UN- ED IFA C T Standards. The elapsed time taken to develop seven draft M IGs was 6 m onths.
The follow ing table describes and com pares m essage developm ent for each project.
Transport - UN-EDIFACT Phase 1 - Data Analysis
Data content o f documents used by various trading partners was identified. The purpose o f data as used by sender and receiver was examined to eliminate redundant or non- essential items. Ambiguities in data and/or its meaning were resolved. Codes were assessed.
Phase 2 -D a ta Flow Analysis
Data origin and flow through the trading chain was documented for different scenarios to ensure data was entered at source. M odelling led to the change of project scope to include booking messages, formerly excluded.
Phase 3 - Data Specification
Information was broken down into individual items for transfer to the data elements of UN-EDIFACT messages.
Phase 4 -Industry Process Review
Existing industry practice was analysed to identify and eliminate unnecessary document exchange (eg use o f waybills rather than Bills o f Lading that require a signature for legal purposes). Opportunities were taken to streamline processes. Sophisticated IT and EDI users raised User awareness o f how technology can reduce cost and increase processing speed within organisations.
Super - XML
Phase 1 - Industry M odelling
Models showing major information flows for each transaction were developed. The information flows were identified by users and modelled by data analysts.
Phase 2 - Assignment o f Responsibilities
The responsibilities o f key roles and activities associated with each transaction were assigned by Users and mapped by data analysts.
Phase 3 - Business Transaction Sequence
Priority for message developm ent was established and agreed by the W orking Group and Council.
Phase 4 - Generic Transaction Development.
All data required to process a transaction was identified. This was assisted by existing legal regulations that specified data for reporting. There was greater variation in the data required by SFAs for business transactions, like contribution payments by employers.
Phase 5 - Flow & Function Specification
Message flows and business function throughout the trading chain was confirmed. Booking messages were designed for multi-use (eg by exporters to book ships, road and rail carriers and by packers to book containers.) Data, context and scenarios for each activity were specified and processing rules agreed.
Phase 6 - MIG Development
EDI experts selected the D94B version of UN-EDIFACT as capable <>l providing the data and functionality identified by User Groups. Data items identified in analysis and specification phases were cross referenced to data elements in the Standard by project managers and confirmed by EDI experts who then generated draft MIGs. The size, complexity and structure meant implications o f data placement within message segments required lengthy discussion at User Groups. The use o f 3 letter tags for segments (eg NAD for name and address; codes to indicate purpose) and separators makes sample messages difficult to read by users. Where structure, data and code sets were inadequate, requests for change were referred to Australia representatives on the UN-EDIFACT Joint Transport Group (JRT) for inclusion in a revised internationally “harm onised” Standard.
Drafts were discussed by a combined group o f Users and developers to resolve technical issues and queries.
Phase 5 - Business Scenario Identification
All variations in scenarios and business lines to be supported were identified and the data recorded. Each user consulted experts in their organisation to confirm data models included all data required for each scenario. The number o f scenarios was rationalised where possible to reduce complexity.
Phase 6 - Technology Specific MIG Development
UML Models and data modelling tools were used by data analysts to prepare draft messages for testing. A Data Dictionary was developed for all messages and loaded into a modelling tool. W here possible all specifications were harm onised with ATO specifications for super reporting and regulatory purposes7 which meant some application systems already com plied with the specified field types, formats and lengths. Drafts were reviewed and revised by SIGs before referral to the W orking G roup for final QA and approval for testing.
XML protocol XSD syntax version 1999 was selected by project Steering C om m ittees (and confirmed at the conclusion o f each stage o f testing) because it:
>• Allowed some validation o f message > Readability by users and ease o f QA > Business flexibility in the internet world > Simplicity o f testing.
> Mutual learning.
Security issues (eg authentication) were challenging, PKI was too expensive. PGP was understood and available.
7 In 2002 ATO Reports are transmitted via flat file through a proprietary gateway or sent on diskette.
Testing draft M essage Implementation Guidelines
T ransport Project
Pilots began when the draft m essages were com plete in M ay 1995. The test objectives were to ensure:
У MIGs were suitable for business needs. У Standardisation and com pleteness o f code lists
У Interconnectivity o f com m unication networks provided by the V AN s У Softw are com pliance with MIGs and other relevant standards У Interoperability o f software.
Prototype software was used for test the MIGs and also to assess issues related to autom ated processing. M IGs and translation tables for the 94B Standard were delivered to vendors o f packaged softw are and in-house developers in May 1995. In June and July 1995, the project m anagers worked with providers of in-house softw are and of packaged solutions to develop or change softw are to com ply with the MIGs. VANs updated translation and mapping facilities to enable exchange in D94B format using Syntax A*.
Pilots took place from 7 July to 20 October, 1995 using 3 softw are packages and 2 in-house systems. Sam ple data allow ed softw are to be tested prior to installation within User sites. M essages were exchanged (peer to peer and hub and spoke) am ong 10 trading partners involved in a variety o f export products. VANs provided mail boxes and transm ission of m essages free of charge to participants during the pilot period. Initially m essages w ere exchanged am ong trading partners using the sam e softw are to ensure the guidelines were satisfactory for business use. M essages were then exchanged:
У am ong trading partners using different softw are to ensure com patibility У trading partners transporting different com m odities via different transport
m odes
У through a variety o f trading chains.
Users and softw are developers reported changes required to the m essages to the User Groups.
Syntax A is based on the telex standard and requires the use o f upper case and uses “ * ", “ +” and as delimiters. Syntax В was developed more recently and uses characters not available in the older standard.
The MIGs were agreed by the U ser G roups and the industry for use in A ustralia and New Zealand, published and “frozen” for 3 years until 1998. T radegate A ustralia m aintains and updates the M IGs in accordance with local needs and changes in international Standards.
Super Project
A separate project Steering C om m ittee was set up com prising participants interested in assessing the m essages required to com plete one transaction type. T est objectives were lo:
> Test the fitness for production of each Standard к Ensure cross dependencies are com plete
> Ensure business rules are com plete
V Test the fitness for production o f each protocol (XM L, XSD; em ail; W inZip for com pression; PGP for security)
> Prepare technology specific guidelines.
Between 8 and 14 trading partners were involved in peer to peer exchanges for each o f the 2 projects. Testing took an average o f 6 m onths over an elapsed time of 12 months. XM L m essages were generally prepared and processed m anually although one payroll provider did upload and dow nload data to their application from existing test databases and one SFA used a translation product. Spreadsheets, tem plates schem as and a parser were used to prepare and process XM L test messages. C onform ance suites were generally used X M L m essage testing was hindered by lack o f tools and changes in X SD syntax versions. Learning and skills transfer related XM L took time. The follow ing 5 stage testing process used was designed to overcom e these lim itations:
> Pretest assessm ent of data and specifications > Exchange of main X M L m essage only
> Initial exchange o f all messages required to com plete transaction
> Exchange of all m essages using business rules and all protocols except security
> Final exchange o f all messages using all protocols and draft guidelines. C hange requests agreed by the Steering C om m ittee, were referred to SIGs at the end o f each phase. Revised M IGs were used in the next test round.
The lack o f a single body accountable for X M L m eans changes to syntax are difficult to achieve even if participants are active on international bodies. A request to W 3C to allow the use o f in m essages was unsuccessful.
A m persand rem ains a reserved character in XM L because o f the notion o f entities and entity references in XM L.
M IGs were frozen for 2 years and m ade available to participants progressively for use by the Superannuation Industry in A ustralia. A perm anent body is to be established by the Super Industry to m anage the trading com m unity and maintain and update MIGs.
M essage Im plem entation
A rollout o f the transport MIGs was undertaken by Tradegate A ustralia. Initial im plem entation was facilitated by the existence o f com pliant EDI enabled softw are and in-house system s used in the pilot project. Tw o participants im m ediately adopted the MIGs for live peer to peer exchange o f 5 m essages in the trans-Tasm an trade. VANs and softw are vendors assisted take-up. W ithin 12 months 2 o f the 7 com m only used packages were com pliant. W hen trading partners agreed dates for com m encing e-transactions, com pliance and interoperability was assessed during testing with the new partner prior to live exchange. The use o f UN-ED IFA CT facilitated international and cross industry exchange (eg with retailers) o f electronic messaging.
Although T radegate A ustralia had experience with EDI im plem entations, and was able to provide technical assistance, education and m arketing, the cost o f system s integration, EDI technology and VANs hindered take-up by sm aller organisations in the trading chain until industry wide use o f IT and internet was increased. W hen a web-based e-Form s solution was im plem ented usage by sm all organisations increased significantly.
In the Super Industry the lack o f appropriate XM L X SD tools and expertise resulted in slow er than expected initial take-up. The first live transaction o f 1 m essage betw een 2 trading partners took place 11 m onths after the M IG was published. Delays resulted from the com plexity o f establishing e-gatew ays, integration with existing application and processing system s and Straight Through Processing. Cost o f im plem enting new technology is alw ays high m aking cost/benefit less viable for early adopters.
Partners were tested independently for com pliance o f their e-gateway system s and m essaging with the M IGs and “certified” to exchange m essage types. By M arch 2002 2 out o f 20 vendors o f softw are and tools were considered com pliant, but a further 6 vendors will be com pliance within 6 m onths. XM L does have the advantage o f facilitating w eb-based solutions, data m apping and the integration and aggregation of data from different sources sim plifying peer to peer exchange. Interoperability with banking was achieved
and com patible XM L m essage developm ent is underw ay w ithin the investm ent industry.
Conclusions
In addition to the issues encountered at each phase in the case studies and the outcom es described, key factors related to the choice o f m essaging relate to the:
>
B usiness requirem ents and priorities o f the different business types within the industry and trading com m unity. For Transport, international and inter industry exchange was essential and U N -ED IFA C T was the established Standard. For Super, the ability to integrate with web interfaces was very important.> T im e for com pletion o f m essage developm ent and im p lem en tation . A m essaging protocol can increase the com plexity o f projects and significantly expand the time required to com plete the project and therefore the risk o f failure.
> Interoperability am ong trading partners. C ertification o f e-gatew ays of trading partners, softw are and other products is required prior to adm ittance to the e-Trading Com m unity. Transport EDI users in 1995 reported continuing difficulties in sending messages to partners using different softw are packages and/or connected to a different VAN due to the use o f proprietary interfaces. This discouraged take-up. All protocols must be interoperable and reliable. All trading partners must adhere to established business rules and industry procedures.
> T ake-up rates and achievem ent of critical m ass. T ake-up is a factor o f cost benefit, availability of solutions, sim plicity, risk, confidence in the technology and m essaging, and peer and com m unity pressure. B enefits are realised when most transactions with key trading partners com ply with industry ecom m erce standards. New protocols m ean new technology and added cost. Business cases are less viable and cost/benefit is less favourable for early adopters. Technical skills may not be available in house and be in short supply in the market.
>
M aturity and stability o f m essaging protocol. T he frequency o f revision and ease o f am ending M IGs are im portant factors affecting system s m aintenance costs. The A ustralian representatives on the U N /ED IFA C T JR T were able to expedite changes to the Standards to ensure they m et the needs o f the Transport project. There is no single body responsible for XM Land there are num erous types o f XM L (eg XDR, XSD, DT D). The selection o f an im m ature, unstable m essaging protocol and syntax before tools and organisational expertise are available does carry a higher risk for project m anagem ent and im plementation. The Super project confronted tools that were not backw ardly com patible, products where APIs w ere not available and com plex technical integration issues that took additional tim e to resolve. In 2002 the selection of XM L is considered a higher risk than UN- EDIFACT.
>
U niversality. In global com m erce most industries require m essaging that is internationally "harm onised” and can be used for e-transactions with trading partners in other countries. If m essages are to be exchanged internationally, and the sender and receiver speak different languages, the tags should be “ language neutral” . This issue is illustrated below.UN-EDIFACT: Em ployer name and address
N A D +B G +EM PLO Y ER ID ++A C M E M A N U FA C TU R IN G +1234M AIN STR E ET+ FA IR FIELD +V IC + 3210’
X M L XSD: Em ployer name and address <Em ployerD etails>
<Em ployerID >A C M E M A N U F A C T U R IN G S Em ployerID > < A ddressL inel> 1234 M AIN STR E ET < /A d d ressL in el> <A ddressLine2> </A ddressLine2>
<Suburb>FA IR FIELD </Suburb> <State>V IC </State>
< P o stco d e> 3 2 10</Postcode> </Em ployerD etails>
On the basis of experience illustrated in the two case studies, in addition to considering the key factors related to the choice o f m essaging, the follow ing criteria should be applied when determ ining the appropriate m essaging protocol for eC om m erce initiative:
> M essage standards are mature > T ools are available and m ature
> Skills and expertise related to the m essaging protocol and the associated technology are available within participant and vendor organisations
r Project elapsed time remains short (less than 18 m onths)
'r Integration with existing with technical and business system s is understood > Successful exam ples are established in sim ilar industries
> Key partners and vendors will com m it to im plem entation if the m essaging protocol is selected
> Interoperability am ong trading partners within the industry and with inter connected industries can be achieved within a realistic tim efram e and at an accepted cost.
References
1. Assenza, F. & Walsh, P. (1995) "ED1MI - EDI fo r M aritime Imports - Final Project Report" Tradegate Australia Limited, Sydney, December 1995
2. Cameron J, & Jeacle P. (1995) "EXTEDI - EDI fo r Maritime Exports - Final Project Report" Tradegate Australia Limited, Sydney, December 1995
3. Cameron J & Clarke R.A. (1996) "Towards a Theoretical Framework f o r Collaborative Electronic Commerce Projects Involving Sm all and M edium-Sized Enterprises" Proc. 9th ED1-IOS Conf'., Bled, Slovenia, June 1996
4. Clarke R.A. (1994a) "EDI in Australian International Trade and Transportation" Proc. 7th EDI-IOS Conf., Bled, Slovenia, June 1994
5. Finkeide, K, Blair, A., Lai, K., and McCarthy, B. "SIW G M essage D evelopm ent Process M ethod 1.1" SIWG, January 1999
6. Westley, F. & Vredenburg, H. (1991) "Strategic Bridging: Collaboration Between Environment a nd Business in M arketing o f Green Products" Journal ot Applied Behavioral Science Vol 27 Issue 1 (May 1991) pp.65-90
