A. White, K. Peterson, B. Lheureux Research Note 1 February 2003 Commentary New P2P Solutions Will Redefine the B2B Supply Chain As multienterprise business processes evolve on a large scale, application architectures based on peer-to-peer computing will emerge. These solutions will have a major impact on the business-to-business supply chain. Peer-to-peer (P2P) computing has been associated with and popularized by file-sharing applications (such as Napster), as well as mainstream services, such as AOL Instant Messenger (IM) and Microsoft MSN Chat. Thus far, however, enterprises have only used client/server architectures to manage supply chain processes, and P2P has not been visualized in the context of business-to-business (B2B) activities. As enterprises seek more opportunities to achieve competitive advantage outside enterprise boundaries with strategic trading partners, innovators will exploit shared business processes that combine enterprises in multienterprise and collaborative-commerce (c-commerce) initiatives, and these processes will be enabled by a P2P computing architecture. Here, we discuss extending the distributed P2P computing model for the supply chain and explain how a "trading grid" will emerge that will facilitate a shift in competitive and economic power in the value chain. Extending the P2P Computing Model In "The Five Peer-to-Peer Models: Toward the New Web," Gartner defined the P2P models. The original model explained how file sharing and resource sharing were supported in a P2P model vs. a client/server approach. Relationships between buyers and sellers are more complex, and they cannot simply be "Napsterized"; however, the basic P2P model assumptions persist. We extend this model to support the supply chain process, which involves the following: Sharing services across a network this includes content (files, documents, P2P file sharing); resources (CPU cycles, disk storage, network bandwidth, grid computing, policy-based computing); and multienterprise business processes Networked nodes this includes client PCs and servers that behave as clients and servers simultaneously (symmetry in network access and service is a characteristic of P2P computing applications) Integration how users connect with each other Gartner 2003 Gartner, Inc. and/or its Affiliates. All Rights Reserved. Reproduction of this publication in any form without prior written permission is forbidden. The information contained herein has been obtained from sources believed to be reliable. Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. Gartner shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof. The reader assumes sole responsibility for the selection of these materials to achieve its intended results. The opinions expressed herein are subject to change without notice.
P2P computing is usually focused on IT resource use, rather than on enterprise business processes; however, P2P computing mirrors major characteristics for how enterprises work internally, as well as with trading partners. Some business processes, such as purchasing, are enterprise-centric. Purchasing integrates with order management to reflect purchasing data in the seller's enterprise applications. Some business processes have a multienterprise focus. One such process, serviced by centralized e- marketplaces, is hosted auctions, in which multiple buyers and sellers congregate virtually. This centralized process replicates enterprise data and also uses the same connectivity technology as singleenterprise-centric business processes. These processes were centralized because some vendors thought that they would be more efficient. P2P computing supplies the application architecture needed to provide multienterprise business processes, using distributed resources where they are, rather than in a centralized fashion. Furthermore, instead of having layers of individual private extranets linked by proprietary connections to each trading partner, a trading grid emerges, where all members of a given industry connect using the same mechanism. This is the "Napsterization" of the supply chain. s can exploit the value of hidden information across the enterprises using the "Service Station" model (see "The Service Station: A P2P Web Services Usage Model"), as well as take advantage of P2P computing for large-scale content sharing. In "Will Grid Computing Create New Service Provider Opportunities," Gartner reviews how a single network of shared CPU resources can change the way application services will be offered. In the public eye, this is typified by the seti@home project, which replicates an identical process in parallel across thousands of separate CPUs. The P2P Service Station Scales to the Trading Grid By leveraging the service station P2P model and overlaying it with the distributed network of buyers and sellers in the supply chain, we arrive at the trading grid, a dynamic, yet deeply rich, collaborative network of trading partners. Connected and connectable with each other, such networks share content and resources, developing in the more strategic relationships large-scale, multienterprise c-commerce business processes. For example, the request for proposal (RFP)/request for quotation (RFQ) response business process (see Figure 1) can be re-implemented and made more efficient and convenient using the service station concept. Although it's not a shared business process, it demonstrates how a business process that is currently being serviced via client/server technology can be deployed using P2P computing. 1 February 2003 2
Figure 1 RFP/RFQ Business Process Based on a P2P Computing Framework Buyer A 3 Response Seller A 1 RFP 5 4 Buyer B 2 Network (Internet) 4 3 Response Seller B Source: Gartner Research The steps that would occur are as follows: 1. Buyer posts RFP to its own server 2. The networked suppliers' remote application "sees" the open RFP (via role-based access) 3. Partner responds to RFP locally (manually or via agent) 4. The networked buyers' remote application "sees" the new responses and accesses (replicates or accesses them remotely) as needed (via role-based access) 5. Responses are ranked, and an order is created in the enterprise application to chosen partner The RFP/RFQ process is more efficient, because the data is not replicated centrally. Based on the adoption of industry standards, this model is more scalable, because partners can connect their distributed data at lower cost. By extending this concept to all buyers and sellers in an industry and looking at all business processes, a trading grid will develop. This sort of network has less data replication and is more resilient, because there are fewer points of failure. Other advantages of the P2P approach for enterprises include the following: Reduced business costs achieved by better and more efficient business and IT resource usage Improved service levels achieved by dynamic adjustment and self-organizing network connectivity, via more-resilient, shared business processes Increased potential for agility achieved through faster alignment of planning and execution of business activities across multiple trading partners directly connected There are many other examples of business processes with characteristics similar to the RFP example. Those characteristics include the following: 1 February 2003 3
Large numbers of participants (buyers and sellers) Known membership buyers and sellers know each other and have an established relationship or can establish it quickly Availability of common, shared, or federated terminology or data standards Common or shared integration framework, assuming security and scalability The necessary technologies will not mature or converge to support scalable P2P in the supply chain until 2010 (0.8 probability). The technologies that are maturing and converging in supply chain management (SCM) during the next few years to enable this P2P computing framework to take shape including the following: application integration (EAI), which is evolving to support transaction delivery and Web services networks (see "EDI to Web Services Networks: Contrasts in Convergence") providing the connections between peer extranets, e-marketplaces and third-party hosted solutions Grid computing and policy-based computing providing the shared space in which CPU cycles are aggregated XML-based document standards based on Extensible Markup Language (XML) for industry and interindustry interoperability (including UCC, RosettaNet, OASIS) proliferating to provide context across networks Atomistic and non-atomistic registry models for the large-scale synchronization of data and processes (for example, UDDI and UCCnet) for the larger-scale ecosystem of multiple networks Ubiquitous forms of authentication and other required security capabilities Through 2006, enterprises should opportunistically select supply chain B2B c-commerce solutions built on P2P computing models for strategic competitive advantage, because the combination of P2P computing in B2B will remain immature. One Flaw in P2P Computing Remains Unresolved The flaw in P2P computing that will hold back its exploitation for B2B is interoperability. This is the ability of different applications to integrate and coexist when not using the same technology in this case, a P2P trading grid. This drawback will limit the scalability of such applications and will keep transaction and switching costs higher than in the futuristic P2P-based trading grid. The lack of interoperability will slow the connectivity between large networks headed by channel masters or brand owners. Most P2P computing services are not interoperable. If users of Grokster could search for music files (such as MP3s) on Kazaa, the two file-sharing services would be said to be interoperable. If users of AOL IM could "chat" with users of Microsoft MSN Chat, those two systems could be called interoperable. Interoperability is the ability to operate as if a user were physically located on the same network. P2P in the supply chain requires a higher level of interoperability as well business process interoperability. This is different from the interoperability required by systems such as enterprise resource planning (ERP), customer relationship management (CRM) and supply chain planning (SCP), because it refers to business process symmetry. This is more than just the trading of XML documents or files, although it is likely that, as new business processes emerge, they will cause a revision in such document standards. Business process interoperability standards are unique to multienterprise business processes. They are required when a process is symmetrical across trading partner boundaries. In enterprise-centric business 1 February 2003 4
processes purchasing integrated with order management, for example the process is not symmetrical. In collaboration, the business processes need to reflect some level of symmetry, so that both partners follow a common or shared process. This can be seen in a few industry initiatives, the most notable being collaborative planning, forecasting and replenishment (CPFR). CPFR is a symmetrical business process not because partners are equal in their efforts, but because the process is mirrored, in whole or in part, by the partners. All business processes that are shared between multiple trading partners will exhibit a level of process symmetry. P2P computing combines the sharing (see Figure 2) of content, resources and business process between users on proprietary and open networks. Figure 2 P2P Computing Stack Interoperable Shared Content Atomistic Search Proprietary Shared Resources Shared Business Processes Non-Atomistic Search Source: Gartner Research By 2008, standards for business process interoperability will emerge as the primary impediment to the mass adoption of large-scale multienterprise B2B and c-commerce initiatives (0.8 probability). Once in place, and assuming Web services standards mature, business process interoperability standards will enable the rapid deployment of multienterprise c-commerce based on the P2P computing architecture. Once standards interoperability is achieved, networks or value chains will be able to connect to each other to create the trading grid. Integration, transaction and switching costs between enterprises and value chains connected to the trading grid will then begin to decrease. The trading grid will not spring up automatically. The first signs of grid formation will be driven by large channel masters and brand owners that, as leaders and owners of their own value chains, will exploit P2P computing and, therefore, create overlapping networks. Over time, midtier enterprises that compete or service more than one channel master or brand owner will be forced to connect to the grid, and so the grid itself will then coalesce. By 2010, the majority of business processes that cross trading-partner boundaries and provide competitive advantage will exploit shared business processes and leverage P2P computing models (0.7 probability). Beyond 2010, solutions initially deployed as simple extended applications for SCM outside the enterprise's four walls will begin to fail when attempting to synchronize large amounts of data across multiple networks, or when more-complex multienterprise business processes are needed. Consequently, application vendors will embed P2P computing application architectures. Bottom Line: Beginning in late 2006, the application architecture used to deploy shared business processes for multienterprise supply chain management (SCM) and c-commerce applications will begin to focus on exploiting a peer-to-peer (P2P) computing platform. These solutions will combine technologies from registry services, distributed database management, grid and policy-based computing, and 1 February 2003 5
collaboration toolsets. Over and above enterprise-centric and cooperative business-to-business (B2B) processes, these P2P-based collaborative business processes will offer the most strategic competitive advantage to their users. 1 February 2003 6