A hybrid web-centric and peer-to-peer architecture for consumer-to-consumer e-commerce.

Otto, James ; Wagner, William ; Pant, Vik 等

ABSTRACT

An exciting area of growth in the world of E-Commerce has been the rapid rise of the Peer-to-Peer (P2P) architecture. While popular, as witnessed by the widespread use of file sharing programs such as KaZaa and Morpheus, this architecture has not yet been leveraged into a profitable business model. This paper proposes a new architecture for Consumer-to-Consumer E-Commerce that combines the flexibility of the P2P architecture, with the more stable web-centric architecture. These two architectures are analyzed from the perspectives of their relative information usefulness and mercantile activities they will support. After a discussion of the tradeoffs of each architecture, a C2C business model that takes advantage of the relative strengths of a hybrid P2P and Web-Centric architecture is presented.

INTRODUCTION

This paper analyzes the tradeoffs between a distributed peer-to-peer (P2P) architecture and a web-centric architecture (the dominant e-commerce architecture) for consumer-to-consumer (C2C) online market makers. The analysis is used to compare the relative utility of both P2P and Web-centric architectures to support C2C transactions from information usefulness and mercantile activities perspectives. The paper then suggests a potential C2C business model based on a hybrid P2P and Web-Centric architecture.

The analysis of architecture and its fit to business processes is important because the alignment of architecture with user processes is relevant to the success of organizational goals (Carleton, 2002; Eisenmann 2002). Since C2C mercantile interactions are fundamentally the transfer of information, goods and services, and payments between consumer peers, it makes sense that P2P based information architectures be considered for C2C interactions and business models.

This paper addresses online market makers - electronic intermediaries that provide market services to participants by providing an infrastructure and place to trade along with rules to govern trading. Specifically, we address the online market maker business model for P2P C2C.

There are numerous business models supporting online market makers that can be categorized by product focus, transaction type, and affiliation as outlined in Table 1 (Eisenmann, 2002). The last column of Table 1 categorizes the scope of this paper.

As shown in Table 1, this paper addresses the market for C2C participants, standardized, horizontal, and indirect products, catalog transactions, with a neutral affiliation. A good example of this type of market is the traditional newspaper classified consumer advertisements for well understood products such as cars or appliances. The classified section of the newspaper provides a forum for catalog information for sellers for a fee. This same type of consumer classified market is available online for web-centric architectures. For example, see Figure 1.

[FIGURE 1 OMITTED]

We chose this market segment because it is less complex than many other markets and it is essentially P2P once the buyer and seller find one another. In other markets, a variety of complex services may be necessary to accompany the transactions. For example, in auction markets, bidding must be tracked over time using relatively complex rules. In B2B markets, a variety of more exotic business support services can be required, such as request for information (RFI) document generation, requisition and routing approval, financial settlement of orders, controls over who can see information, assurance that transactions are in accordance with regulations, laws, and pre-negotiated contracts, etc. (Katerattanakul, 2002; Smaros, 2001).

For our tradeoff analysis, we assume that the P2P architecture is not centrally controlled or managed whereas the Web-centric architecture institutes centralized control over user information and activities. We also assume that C2C sellers have an always-on connection (such as Digital Subscriber Line (DSL) or cable modem), as opposed to a dial-up connection. If the user does not possess an always-on connection, then the analysis becomes trivial because the 24/7 availability of a Web server dominates any tradeoffs.

OVERVIEW OF P2P AND WEB CENTRIC ARCHITECTURES

According to WhatIs.com, "...peer-to-peer is a communications model in which each party has the same capabilities and either party can initiate a communication session." (TechTarget, 2003). This means that there is no intermediary between the parties (except for their internet service providers). For P2P connectivity to function on the Internet, users connect to each other via the other party's IP address. Since every device (computer, PDA, etc.) or service (routing, bridging, etc.) connected to the Internet has a unique IP address, it (IP address) is very helpful in identifying and locating these devices/services. As displayed in Figure 2, since both parties in a P2P connection have the capability of being clients and servers, they are termed "hosts" or "nodes".

[FIGURE 2 OMITTED]

It is the decision of the user to provide services or consume them. If the user chooses to be a provider then he or she can determine limits on services/resources that can be used. These limits can be based on a variety of factors such as security privilege and usage history of connected peer, and also on system load, bandwidth usage, network throughput, time of day, etc.

A Web-Centric Architecture is a type of client/server architecture. In such an architecture, one (or many) machine(s) produce and provide the services and one (or many) machine(s) acquire and consume these services. According to Webopedia.com, client/server architecture is, "a network architecture in which each computer or process on the network is either a client or a server. Servers are powerful computers or processes dedicated to managing disk drives (file servers), printers (print servers), or network traffic (network servers). Clients are PCs or workstations on which users run applications. Clients rely on servers for resources, such as files, devices, and even processing power" (Webopedia.com, 2003). A Web-Centric architecture is the architecture that defines the World Wide Web. In such an architecture, there is a clear demarcation between the producer and the consumer of the services that interact with each other over the Internet. In a Web-Centric architecture, some computers are dedicated to providing services to others. In most cases, applications implemented over the client/server architecture, such as WWW websites, are meant to serve many clients at the same time.

ANALYSIS OF PEER-TO-PEER AND WEB-CENTRIC ARCHITECTURE TRADEOFFS

Our framework contrasts the general architectural tradeoffs of P2P and Web-centric (client-server) architectures. We recognize that these tradeoffs will vary depending upon the specific application context, so we provide an explanation of how these tradeoffs apply. Using this information, the reader can adjust the analysis to fit their specific context.

Alter (2002) has proposed a series of criteria for evaluating E-Commerce models using an information perspective. These are summarized in Table 2 and are used as a basis for comparing the relative advantages of the P2P and Web-centric architectures from an e-commerce consumer perspective. This evaluation also assumes that the user interface criteria (such as usability, user friendliness, responsiveness - etc.) can be made equivalent between the two architectures by proper design of the user interface layer. Since the transactions between buyer and seller over the Internet largely involve information exchange, the criteria are discussed in information terms.

Information Accuracy: This criterion measures how well the information represents what it is supposed to represent. In centralized systems, a single master copy of the data can be maintained and enforced. Any replication is performed explicitly (usually for backup purposes), so there is little chance that alternate inconsistent files exist. In a distributed P2P system, the same file can exist in several different places. For example, P2P users can download a product information file from the originator. If that downloaded file is modified, then two different files will exist that purportedly refer to the same product--which can breed inconsistencies and errors (Seltzer, 2002; Killdara Corp, 2001). Thus for this measure, the Web-centric architecture has the advantage. Additionally, because information is centrally monitored and controlled in a Web-centric system, inaccurate data can be removed from the system. In a truly distributed system, no such enforcement mechanism exists.

Information Precision: This measure addresses how well the information detail meets user requirements. In a Web-centric architecture that is centrally controlled, the user is restricted to supplying information in the format prescribed by the management authority. This may not allow for the level of information precision desired by the users. For example, a user may wish to provide a high resolution image of a product that exceeds the allowed web server file size. On the other hand, a distributed P2P system allows any desired data to be presented by the user, from videos of the product to detailed images with narration. The buyer and seller select the amount of information, along with the associated bandwidth and storage requirements, that best meet their transaction needs. Thus for this measure, the distributed P2P architecture has a relative advantage over the centralized web architecture.

Information Age: This criterion concerns how old the information is relative to the requirements of the task. In a Web-centric system, information can be controlled and removed from the system by centralized management when it expires (such as when the client sells the product). In a distributed P2P system, the architecture places that responsibility on each user. If users do not remove information from the network, then old, irrelevant information may still be available to users. Thus, the Web-centric architecture has an advantage for this measure.

Information Completeness: This criterion addresses how adequate the amount of information is for the task at hand. The Web-centric architecture, which is centrally managed, can limit the amount and format of information provided by the seller or buyer. The P2P architecture only limits information to that which can be handled by the bandwidth, storage, and format constraints of the buyers and sellers. Thus, the P2P architecture can provide better information completeness due to its scalability (Smaros & Framling, 2001). As an example, the amount of information stored and available to users of KaZaA, a popular P2P application is huge. For example, at 8pm on 16 March, 2004 the following statistics applied to Kazaa:

Over 2.5 million users online

Over 1.4 billion files

Over 4 petabytes (4 million gigabytes) of information

Information formats included audio, video, images, documents, software, and music playlists

Information Timeliness: This measures whether the information is provided quickly enough to meet user needs. In the P2P architecture, the information is available immediately (Merkow, 2000) as soon as the user moves the files into the shared folder. For the Web-centric system, the user must explicitly communicate information to the administrators (Smaros & Framling, 2001) and may need to wait until the user and/or information is approved for posting, which may reduce the timeliness of the data. Therefore, the P2P architecture has an advantage for information timeliness.

Information Source: This measure addresses the credentials of the supplier of the information. Is the source credible, trustworthy, and free from bias? Web-centric systems, with their centralized control, can institute trust mechanisms to verify system users. For example, Ebay provides various utilities for providing feedback on sellers and also for filing more formal complaints. These same types of verification methods are not as robust in a distributed P2P system. Thus, the Web-centric architecture has an advantage for this measure.

Information Availability: This measure addresses the extent to which the necessary information is available to the people that need it when it is required. In practical e-commerce terms, it is expected that data will be available 24 hours per day, 7 days per week (24/7). Web-centric systems can explicitly manage the replication of data and system redundancies to cope with a failure or error. In a P2P system, redundancy can, but does not necessarily, occur. For example files may be copied by prospective buyers, but this may not necessarily happen. Thus, if the file is copied by many users, it is possible that redundancy could be quite good (at the possible expense of information consistency). However, since there is no centralized management, replication of files is not explicitly required. Thus, it is also possible that redundant files may not exist in a P2P system. In a centrally controlled system, redundancy can be enforced. Additionally, 24/7 availability requires that the information source be continuously operating. A centralized web server has people dedicated to keeping the web server operating. This level of reliability may not be available to your average P2P user (Note that this analysis assumes that the P2P users have an always-on internet connection). For these reasons, the Web-centric approach has an advantage over P2P when it comes to information availability.

Information Admissibility: This criterion focuses on whether the information provided is socially, culturally, or legally appropriate. In a Web-centric system, management can remove objectionable information, whereas this does not occur in a distributed P2P system. For example, Ebay management removed items by individuals attempting to sell debris from the Space Shuttle accident. This problem with P2P has also been heightened by the wide press coverage of P2P sites such as Napster, which may be involved in intellectual property thefts (Carleton, 2002). Thus, the Web-centric system has an advantage for this criterion (unless the seller or buyer are specifically interested in exchanging inappropriate information).

Information Expense: This measure deals with the cost of executing transactions. In the P2P architecture, client software interacts directly with other client software. The expenses associated with data storage and bandwidth are borne by the client. With a Web-centric architecture, these expenses are borne by the manager of the Website. It is assumed that these bandwidth and storage costs, plus any profit, are then passed on to the users as fees (Smaros & Framling, 2001). If one assumes that a significant portion of user bandwidth and storage stands unused and idle, then the marginal costs of adding the P2P functionality will be small. Thus, from a user perspective the P2P architecture has an advantage.

Information Security: This measure addresses the level of protection against unauthorized information access or alteration. Centrally managed security can be more attentive and sophisticated than security that relies on each user. It is well recognized that security is the Achilles heel of P2P applications because of the average user's limited awareness of security issues (Kalakota, Susarla, & Parameswaran, 2001). The Web-centric approach has an advantage over P2P which has more places where security can be breached (Killdara Corp, 2001).

Information Presentation: This measure addresses how the information is presented and how appropriate it is for the desired application. Since P2P allows users to provide information in any format desired (Framling & Holstrom, 2000), it has an advantage of the centrally controlled Web-centric architecture.

Information Standardization: This criterion measures how consistent information is across similar files. Standardization can help users because it supports a consistent view of information and easier information exchange. Since the Web-centric architecture supports the centralized management of standardization and can also provide translation services between different information formats (Smaros & Framling, 2001), it has an advantage over the P2P architecture.

C2C TRANSACTIONS AND SUPPORTING E-COMMERCE SERVICES

In this section we compare the P2P and Web-centric architectures from the perspective of a consumer mercantile model. In C2C e-commerce, the mercantile activities of the buyer and seller are listed below (Kalakota & Whinston, 1996).

* Product/service search

* Comparison shopping and selection

* Negotiation of terms

* Placement of order

* Authorization of payment

* Receipt of product

* Customer service and support

Table 3 compares the ability of the P2P and Web-centric architectures to support e-commerce implementations of each of the consumer mercantile activities. For each mercantile activity, Table 3 lists the activities from both the buyer and seller perspectives and the electronic implementations of services that support those mercantile activities.

For example, during the negotiation phase, Table 3 shows that during the product search phase, the buyer can search web pages and web catalogs in a Web-Centric architecture or specialized P2P search engines in the P2P environment. Additionally, the buyer can review all the offerings provided by a specific peer site by reviewing all the files in the seller's shared folder.

A PROPOSED HYBRID ARCHITECTURE

If one examines the relative advantages and disadvantages of Web-centric and P2P architectures in Table 2, and their respective services in Table 3, it is evident that the architectures have different strengths and weaknesses. Given this, we propose that a hybrid architecture can best meet the needs of the C2C buyers and sellers. Figure 3 diagrams such an architecture. This hybrid architecture allows one to place C2C services at either the peer clients or the web server depending upon the nature of the service.

[FIGURE 3 OMITTED]

To show the relative utility of this hybrid model we can apply the same information-centric perspective suggested by Alter (Alter, 2002) that was discussed earlier. Table 4 summarizes the various services that could be provided by the two parts of the hybrid model.

Kwok, et. al. provide a three layer framework for P2P business and service models (Kwok, Lang, & Tam, 2002) as diagrammed in Figure 4. The bottom two layers define the technology and community service application layers. These bottom layers provide the infrastructure to support the business model of the top layer. It could be argued that the first two layers of the Kwok model are being successfully addressed by some current P2P applications given the large user participation levels. For example, a survey of Cornell student usage of bandwidth found that on average 55% of the bandwidth was used by Kazaa inbound and outbound traffic (Cornell University, 2001). This is far more than any other type of activity on campus.

[FIGURE 4 OMITTED]

Assuming that the infrastructure and first two layers are being adequately addressed by current P2P communities, then it may be possible to develop the revenue to support a business model layer by providing a combination of free and paid value-added services. For example, one could place those services that require unbiased third party and/or centralized control at the web server. Bandwidth intensive, storage intensive, and rapidly changing information could be placed at the P2P client locations. Basic services such as search and referral services could be provided free of charge, while premium support such as escrow services could be provided for a fee. Table 5 summarizes ideas about how free and charged services could be split between the P2P and Web-Centric portions of the architecture.

SUMMARY AND CONCLUSIONS

This paper has examined the relative advantages and disadvantages of the P2P and Web-Centric architectures for C2C transactions. Based on this analysis, a hybrid architecture was proposed along with suggestions on how the hybrid P2P and Web-Centric architecture might be leveraged to support a profitable business model.

REFERENCES

Alter, S. (2002). Information Systems: The Foundation of E-Business. Upper Saddle River, NJ: Prentice Hall.

Carleton, J. (2002, May). Fast growth forecast for enterprise P2P. Tech Update. Retrieved August 17, 2004, from http://techupdate.zdnet.com/techupdate/stories/main/ 0,14179,2863979,00.html

Cornell University (2002). Cornell Internet Usage Statistics for 2001. Retrieved August 15, 2004, from http://www.cit.cornell.edu/computer/students/ bandwidth/charts.html

Eisenmann, T. (2002). Internet Business Models: Text and Cases. New York, NY: McGraw-Hill.

Framling, K. & J. Holmstrom (2000). A distributed software for collaborative sales forecasting. Second Conference on Management and Control of Production and Logistics, July 2000.

Kalakota, R. & A. Whinston (1996). Electronic Commerce-A Managers Guide. Reading, MA: Addison-Wesley.

Katerattanakul, P. (2002). Framework of effective web site design for business-to-consumer Internet commerce. INFOR, 40(1), 57-70.

Kazaa (2004). Statistics accessed from Kazaa peer-to-peer network on August 16, 2004 using the Kazaa client P2P software.

Killdara Corporation (2001). Peer-to-peer: Good, bad and in-between. Retrieved August 11, 2003, from http://www.killdara.com/whitepapers.htm

Kwok, S., Lang, K. & K. Tam. (2002). Peer-to-peer technology business and service models: risks and opportunities. Electronic Markets, 12(3), 175-183.

Merkow, M. (2000, October). What can P2P do for B2B. Insights--EC Outlook. Retrieved Aug 17, 2004 from http://ecommerce.internet.com/news/insights/outlook/article/ 0,,10535_486031,00.html

Parameswaran, M., A. Susarla, & A. Whinston (2001). P2P networking: An information-sharing alternative. IEEE Computer, 34(7), 31-38.

Seltzer, L. (2002, January). Peer-to-peer: My favorite stupid fad of 2001. Retrieved August 15, 2004, from http://techupdate.zdnet.com/techupdate/stories/main/0,14179, 2837666,00.html

Smaros, J. & K. Framling (2001). Peer-to-peer information systems--an enabler of collaborative planning, forecasting and replenishment. Logistics Research Network 6th Annual Conference, Cardiff, September, 2001.

TechTarget (2004, February). What Is: Peer-to-peer. Retrieved August 15, 2004, from http://searchnetworking.techtarget.com/sDefinition/0,, sid7_gci212769,00.html

Webopedia.com (2004, May). Webopedia: Client/server architecture. Retrieved August 17, 2004, from http://www.webopedia.com/TERM/C/ client_server_architecture.html

James Otto, Towson University

William Wagner, Villanova University

Vik Pant, Ixos Software, AG.

Table 1. Online Market Maker Business Models

Category Characteristics Scope of this paper

Participants Consumers (C): Consumers Consumer-to-Consume
 represents one or more of the r Participants
 transaction participants
 (e.g., B2C, C2C, G2C, etc.)

 Business (B): A business entity
 represents one or more of the
 transaction participants
 (e.g., B2B, B2C, B2G, etc.)

 Government (G): A governmental
 entity represents one or more of
 the transaction participants
 (e.g., G2B, G2C, G2G, etc.)

Product Customized: Nonstandard products Standardized Products
Focus that require a significant
 amount of communication to
 satisfy buyer

 Standardized: Well understood
 products that do not require
 extensive discussions

Market Horizontal: Diverse customer Horizontal Markets
Paradigm base purchase the products

 Vertical: products restricted to
 a specific buyer pool

Product Direct: Items incorporated into Indirect Products
Kind a product

 Indirect: Items used to support
 activities

Transaction Catalog: Lists of standardized Catalog Transactions
Type products from multiple sources

 Auctions: Buyers (or sellers)
 bid for products

 Exchanges: Continuous auctions
 for standardized products with
 fluctuating supply, demand, and
 price

Affiliation Buyer: Biased exchange terms Neutral Affiliation
 towards buyers.

 Seller: Biased exchange terms
 towards sellers

 Neutral: Unbiased exchange
 terms.

Table 2. Evaluation Criteria from an Information Perspective

 Web-Centric P2P
Information Quality
 Accuracy +
 Precision +
 Age +
 Completeness +
 Timeliness +
 Source +
Information Availability
 Availability +
 Admissibility +
Information Expense +
Information Security +
Information Presentation +
Information Standardization +

Table 3. Mercantile Activities Electronic Services

 C2C Transaction

 EC Services
Buyer Seller Provided Web-Centric

Product Provide Info Electronic Website
Search Product Search
 Search Engines

 Electronic Website
 Product catalog
 Advertising
 (Push, Pull)

Compare Provide Online Reviews Website user
Alternatives Comparison and ratings groups
 Information Comparison Website
 Charts reviews
 Seller Database Seller
 ratings
 database

Negotiation Negotiation Communication Terms
 Channels information
 (email, voice Tax/Fee
 chat, Calculators
 text chat) Information
 Terms Web chat
 Information
 Tax/fee
 Calculation

Provide Accept Electronic Credit Online credit
Payment Payment cards

 Electronic Debit Online
 payment
 system
 (e.g., paypal)

 Escrow Services

Accept Fulfillment Shipping
Shipment Information

Request Provide Knowledge Base Online
Service Service Electronic Knowledge
and Support and Support Communication Base, FAQ
 Web Page
 Email, Chat

 C2C Transaction

 Other Internet
Buyer Peer-to-Peer Tools

Product P2P Search Web Search
Search Engines Engines

 Peer catalogs Email
 (solicited and
 unsolicited)
 Newsgroups

Compare Usenet
Alternatives Newsgroup
 reviews
 Web periodical
 reviews

Negotiation Terms Email
 Information
 Tax/Fee
 Information
 P2P Chat

Provide
Payment

 Online
 payment
 system
 (e.g., paypal)

 Electronic
 Escrow

Accept
Shipment

Request Email, Chat
Service
and Support

Table 4. Services provided by Hybrid Model

 Web Server Services P2P Services

Information Quality

Accuracy Provide summary Provide accurate
 information and refer information
 to more accurate peer
 client information

Precision Provide summary Provide precise
 information and refer information
 to more precise peer
 client information

Age Provide valid Provide valid
 timeframe of peer timeframe of peer
 information information

Completeness Provide summary Provide complete
 information and refer information
 to more complete peer
 client information

Timeliness Provide summary Provide timely
 information and refer information
 to more timely peer
 client information

Source Provide ratings/ Provide ratings/
 reviews of seller reviews of seller

Information Availability

Availability Provide 24/7 summary Provide 24/7 detailed
 information information

Admissibility Verify that Provide admissible
 information is information
 admissible

Information Expense Provide summary Provide detailed
 information and information and refer
 reference to peer to Web server for
 clients for free; additional value-added
 charge for additional services
 value-added services

Information Security Provide security Refer to Web-Server
 patches and tips to for security patches
 users and tips to users

Information Provide summary Present information in
Presentation information and refer variety of appropriate
 to peer client formats
 information

Information Provide summary Provide information in
Standardization information in standard and other
 standard format and appropriate formats
 refer to additional
 peer client
 information

Table 5. Consumer-to-Consumer Services on the Hybrid Architecture

 P2P Client Web-Server

Basic Free Information search and Web-Server services
Services presentation capability Summary information with
 Referrals to value-added aging information and
 search
 P2P Chat capability Referrals to P2P site
 information
 Seller ratings
Value-Added Advertising Electronic credit/debit
Pay Services payment acceptance
 Escrow services
 Knowledge base
 Comparison charts
 Advertising