Internet Engineering Task Force O. Kassinen Internet-Draft T. Koskela Intended status: Informational E. Harjula Expires: September 2, 2010 M. Ylianttila University of Oulu March 1, 2010 Taxonomy for P2P Group Management Solutions draft-kassinen-p2psip-group-taxonomy-00.txt Abstract We analyze existing or proposed P2P group management solutions. Based on the analysis, we present a comprehensive taxonomy for the solutions. The highest level of the taxonomoy classifies the P2P group management solutions according to their motivation, group- forming criteria, methods for applying the criteria, and technical realization. Status of this Memo This Internet-Draft is submitted to IETF in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on September 2, 2010. Copyright Notice Copyright (c) 2010 IETF Trust and the persons identified as the document authors. All rights reserved. Kassinen, et al. Expires September 2, 2010 [Page 1] Internet-Draft P2P Group Management Taxonomy March 2010 This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 2. Existing Example Solutions for P2P Group Management . . . . . 3 2.1. Solution 1: "HIERAS: A DHT Based Hierarchical P2P Routing Algorithm" . . . . . . . . . . . . . . . . . . . . 3 2.2. Solution 2: "Trust-Based Community Formation in Peer-to-Peer File Sharing Networks" . . . . . . . . . . . 4 2.3. Solution 3: "Service-Driven Group Management for Mobile P2P Services" . . . . . . . . . . . . . . . . . . . 4 2.4. Solution 4: "A Utility-Aware Middleware Architecture for Decentralized Group Communication Applications" . . . 4 2.5. Solution 5: "PP-COSE: A P2P Community Search Scheme" . . . 4 2.6. Solution 6: "An Interest Group Model for Content Location in Peer-to-Peer Systems" . . . . . . . . . . . . 5 3. The Taxonomy . . . . . . . . . . . . . . . . . . . . . . . . . 5 3.1. The Four High-Level Properties . . . . . . . . . . . . . . 5 3.2. Categories under the Four Properties . . . . . . . . . . . 6 3.2.1. Categories of Motivation . . . . . . . . . . . . . . . 7 3.2.2. Categories of Criteria . . . . . . . . . . . . . . . . 7 3.2.3. Categories of Methods . . . . . . . . . . . . . . . . 7 3.2.4. Categories of Realization . . . . . . . . . . . . . . 8 4. Using the Taxonomy for the Classification of the Example Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . 9 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 6. Security Considerations . . . . . . . . . . . . . . . . . . . 10 7. References . . . . . . . . . . . . . . . . . . . . . . . . . . 10 7.1. Normative References . . . . . . . . . . . . . . . . . . . 10 7.2. Informative References . . . . . . . . . . . . . . . . . . 11 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 11 Kassinen, et al. Expires September 2, 2010 [Page 2] Internet-Draft P2P Group Management Taxonomy March 2010 1. Introduction Several "group" management solutions exist, or have been proposed, for peer-to-peer (P2P) networking. We have done (in 2010) a literature review of more than 50 publications about P2P group management solutions. The literature review will be published in a separate research article. In this Internet-Draft, o we select from those P2P group management solutions a subset that attempts to represent well the versatility of the solutions; and o we propose a simple taxonomy that covers well the P2P group management solutions (both the selected subset and the entire body of the 50+ reviewed solutions), facilitating the classification of the solutions and providing an easy way to overview the properties of the existing or proposed solutions. 1.1. Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119]. Community: A term for a group, emphasizing that the group's members are humans (or the personal network access devices of the humans) who are somehow related and can thus be considered a community. Group: A group of nodes within a P2P network. There are many different definitions for a group in the various existing publications. 2. Existing Example Solutions for P2P Group Management In this section we provide short descriptions about the six selected P2P group management solutions. The six solutions are original inventions of different authors. 2.1. Solution 1: "HIERAS: A DHT Based Hierarchical P2P Routing Algorithm" In [Paper 1], in a system called HIERAS, the nodes in the P2P network are grouped into several Distributed Hash Table (DHT) based overlay networks. Several lower level overlays are created in addition to the highest level overlay. The average link latency between nodes is Kassinen, et al. Expires September 2, 2010 [Page 3] Internet-Draft P2P Group Management Taxonomy March 2010 much lower in the lower level overlays than in the highest level overlay, and this is taken into account when selecting nodes to a group. Routing actions are first executed in the lower level overlays. The goal is to improve routing performance. 2.2. Solution 2: "Trust-Based Community Formation in Peer-to-Peer File Sharing Networks" In [Paper 2], a mechanism for creating P2P groups for sharing academic papers is proposed. The system creates (software) agents, which keep track of the similarity of the ratings of the papers and form trusted agent groups in the P2P network. Agents represent their (human) users; the system forms interest-based communities between the users. There are trust measures for individual agents and groups of agents. Each agent must compute the trust in the agents that it interacts with. 2.3. Solution 3: "Service-Driven Group Management for Mobile P2P Services" In [Paper 3], a peer group management solution based on service semantics and resource availability is proposed. The system is designed for mobile peers. The system involves a centralized server, which holds a service schema influencing the way how peers are dynamically grouped; super peers, which are linked together by the server and act as routing and data-advertisement hubs; and ordinary peers, which hold the actual shared content. The number of overlay network levels is determined by service semantics. 2.4. Solution 4: "A Utility-Aware Middleware Architecture for Decentralized Group Communication Applications" In [Paper 4], a utility-aware middleware architecture called GroupCast for scalable P2P communications is presented. A utility function is presented for quantifying the role of unicast links in the network. A utility-aware distributed spanning tree construction algorithm is presented for implementing efficient message dissemination in group communication, aiming optimizing the utility values of the spanning trees. Also a utility-aware low-diameter unstructured-overlay construction protocol is outlined. The objective of utility-aware overlay construction is that in the resulting network, the neighbors of a node have a high utility with respect to the node. 2.5. Solution 5: "PP-COSE: A P2P Community Search Scheme" In [Paper 5], a system called PP-COSE is introduced for making peer searching efficient. The P2P system is formed by clusters. For each Kassinen, et al. Expires September 2, 2010 [Page 4] Internet-Draft P2P Group Management Taxonomy March 2010 cluster, there exists a node called the search router. The nodes of a cluster register to the search router. The search routers act as community managers and provide message routing. Communities in the system are a different concept from the clusters: community membership is not bound to cluster membership. A filter function exists to reduce the amount of overhead traffic: unpopular communities are filtered out. Clustering can be based on organizational or geographical areas. 2.6. Solution 6: "An Interest Group Model for Content Location in Peer- to-Peer Systems" In [Paper 6], a probabilistic decision tree is used for representing the users' interests. Based on the tree, peers with common interests are organized into the same group. Probabilistic decision trees are used for representing a user's interests, because the interests typically change over time. The system uses a standard probability tree accepted by a majority of users instead of creating a new tree for each user. The aim is to make searches more efficient in unstructured P2P networks. When searching for content, the query is first propagated in the interest group, and only if the content is not found there, then the query is flooded outside the group. 3. The Taxonomy Based on the analysis of the 50+ existing or proposed P2P group management solutions, we present a taxonomy that is suitable for classifying the solutions on a suitable level of detail. 3.1. The Four High-Level Properties In the presented taxonomy, the following four high-level properties are distinguished for a given P2P group management solution. o Motivation: The reason why groups are formed. o Criteria: The criteria that are used for selecting nodes to a specific group. The criteria are, of course, dependent on the motivation. o Methods: The algorihms or other methods for observing and evaluating the criteria, leading to the selection of nodes to a specific group. o Realization: The kind of underlying P2P network that is used as a basis for the solution and enables the running of the specified methods. Kassinen, et al. Expires September 2, 2010 [Page 5] Internet-Draft P2P Group Management Taxonomy March 2010 All four high-level properties should be identified for a given P2P group management solution. However, some properties might not be in the focus of a given solution (i.e., are not presented in the publication), so only a sub-set of the four properties might be defined. In some cases, a publication may also state that some aspect of a solution can be implemented with several different means; this applies usually to the properties "Method" and "Realization". The relationship between the high-level properties is illustrated in Figure 1. +-----------------------+ | Motivation | +-----------------------+ | defines | v +-----------------------+ | Criteria | +-----------------------+ | are evaluated with | v +-----------------------+ | Methods | +-----------------------+ | work on top of | v +-----------------------+ | Realization | +-----------------------+ Classification. Figure 1 3.2. Categories under the Four Properties In this section, we list the different categories that we have observed to exist under the four high-level properties. A given P2P group management solution may exhibit the characteristics of one or more categories within a high-level property. Kassinen, et al. Expires September 2, 2010 [Page 6] Internet-Draft P2P Group Management Taxonomy March 2010 3.2.1. Categories of Motivation o Search Efficiency: The efficiency of routing in the P2P network or (a more specific case) the efficiency of resource searching. o Group Communication: The ability to communicate in a suitable group (of human users). o Service Provisioning: Publishing and discovery of services on top of a P2P system. o Knowledge Sharing and Collaboration: The ability to share important information and collaborate in the context of specific tasks, often in a professional setting. o Trust, Security and Privacy Management: Enhance the trust, security, and/or privacy in P2P operations. 3.2.2. Categories of Criteria o Common Interests: The members (humans) of a community share some common interests, related to e.g. content. o Node Capability: The terminal devices' capabilities such as CPU power or memory. o Level of Trust: How strong is the trust between the users (or nodes). o Social and Organizational Memberships: Membership in a social group or organization. o Locality: * Physical: How near the nodes are to each other, in terms of network latency or other physical network-related metric. * Logical: How near the nodes are to each other, in terms of hop- count or other logical network-related metric. * Geographical: How near the nodes are to each other, in terms of geographical distance. 3.2.3. Categories of Methods o Ontology Matching: There exists a suitable ontology, which is used for finding similarities between nodes. Kassinen, et al. Expires September 2, 2010 [Page 7] Internet-Draft P2P Group Management Taxonomy March 2010 o Link Analysis: Based on the observed interaction between nodes, patterns of interaction are identified. o Attribute Comparison: The non-technical attributes of nodes are compared for finding similarities between nodes. These attributes might include, for example, interests. o Statistical Comparison: Some characteristics of nodes are compared with a statistical approach in order to find similarities. o Analysis of Node and Network Performance: The technical attributes of nodes or the used network are compared, usually for making the network operations more efficient, for example, by minimizing latencies. o Usage of Membership Credentials: Digital credentials are delivered to the members of a group and checked when needed. 3.2.4. Categories of Realization o Single Overlay: The group or groups are created within a single overlay; the entire system contains only one overlay network. * Structured (DHT): The organization and operation of the overlay is based on mathematical rules, usually on Distributed Hash Tables (DHT). * Unstructured: The organization and operation of the overlay is based on some "less exact" rules. + Pure: The system is completely flat (example: Gnutella 0.4), and each node has equal responsibilities. + Hierarchical: There is some structuring, such as edge-peer vs. super-peer distinction (example: JXTA). o Multi-overlay: The group or groups are created using multiple overlays; the entire system contains multiple overlay networks that can be either structured, unstructured, or both. Multi- overlay-based solutions can be further classified to: * Vertical: A vertical system is usually described as a tree, where every layer or leaf is an independent DHT overlay network. In the vertical approach, some nodes usually act as gatekeepers routing the messages between the different levels in the DHT hierarchy. Kassinen, et al. Expires September 2, 2010 [Page 8] Internet-Draft P2P Group Management Taxonomy March 2010 * Horizontal: All the leaf overlays are connected to a single common DHT overlay that is responsible for optimizing the routing in the whole network. Furthermore, the DHT overlay also possesses the functionality and policies how to control the hierarchy of overlays. The different leaf overlays can either be structured or unstructured P2P networks. 4. Using the Taxonomy for the Classification of the Example Solutions In this section, we apply the taxonomy to the example P2P group management solutions that we discussed earlier. o The solution in [Paper 1] has the properties: * Motivation: Search Efficiency. * Criteria: Locality. * Methods: Analysis of Node and Network Performance. * Realization: Multi-overlay, DHT. o The solution in [Paper 2] has the properties: * Motivation: Knowledge Sharing and Collaboration; Trust, Security and Privacy Management (more specifically: Trust Management). * Criteria: Common Interests; Level of Trust. * Methods: Attribute Comparison. * Realization: Unstructured, Hierarchical. o The solution in [Paper 3] has the properties: * Motivation: Service Provisioning. * Criteria: Common Interests; Node Capability. * Methods: Ontology-matching. * Realization: Multi-overlay. o The solution in [Paper 4] has the properties: Kassinen, et al. Expires September 2, 2010 [Page 9] Internet-Draft P2P Group Management Taxonomy March 2010 * Motivation: Group Communication. * Criteria: Locality; Node Capability. * Methods: Analysis of Node and Network Performance. * Realization: Unstructured. o The solution in [Paper 5] has the properties: * Motivation: Search Efficiency. * Criteria: Social and Organizational Memberships; Locality. * Methods: Analysis of Node and Network Performance; Usage of Membership Credentials. * Realization: Unstructured, Hierarchical. o The solution in [Paper 6] has the properties: * Motivation: Search Efficiency; Knowledge Sharing and Collaboration. * Criteria: Common Interests. * Methods: Statistical Comparison. * Realization: Unstructured. 5. IANA Considerations This memo includes no request to IANA. 6. Security Considerations This draft does not introduce any new security issues. 7. References 7.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. Kassinen, et al. Expires September 2, 2010 [Page 10] Internet-Draft P2P Group Management Taxonomy March 2010 7.2. Informative References [Paper 1] Xu, Z., Min, R., and Y. Hu, "HIERAS: A DHT Based Hierarchical P2P Routing Algorithm", 2003 International Conference on Parallel Processing (ICPP 2003), October 2003. [Paper 2] Wang, Y. and J. Vassileva, "Trust-Based Community Formation in Peer-to-Peer File Sharing Networks", 2004 IEEE/WIC/ACM International Conference on Web Intelligence (WI 2004), September 2004. [Paper 3] Liotta, A., Ballette, M., Lin, L., Gasparoni, M., Brick, P., and N. Papadoglou, "Service-Driven Group Management for Mobile P2P Services", Intelligence in Communication Systems, Vol. 190/2005, pp. 221--230, October 2005. [Paper 4] Zhang, J., Liu, L., Ramaswamy, L., Zhang, G., and C. Pu, "A Utility-Aware Middleware Architecture for Decentralized Group Communication Applications", ACM/IFIP/USENIX 2007 International Conference on Middleware (Middleware 2007), November 2007. [Paper 5] Mei, H. and S. Chang, "PP-COSE: A P2P Community Search Scheme", Fourth International Conference on Computer and Information Technology (CIT 2004), September 2004. [Paper 6] Xue, G., You, J., and Z. Jia, "An Interest Group Model for Content Location in Peer-to-Peer Systems", IEEE International Conference on E-Commerce Technology for Dynamic E-Business (CEC-East 2004), September 2004. Authors' Addresses Otso Kassinen University of Oulu Oulu, Finland Phone: +358 8 553 1011 Email: firstname.lastname@ee.oulu.fi Kassinen, et al. Expires September 2, 2010 [Page 11] Internet-Draft P2P Group Management Taxonomy March 2010 Timo Koskela University of Oulu Oulu, Finland Phone: +358 8 553 1011 Email: firstname.lastname@ee.oulu.fi Erkki Harjula University of Oulu Oulu, Finland Phone: +358 8 553 1011 Email: firstname.lastname@ee.oulu.fi Mika Ylianttila University of Oulu Oulu, Finland Phone: +358 8 553 1011 Email: firstname.lastname@ee.oulu.fi Kassinen, et al. Expires September 2, 2010 [Page 12]