Internet Draft I. Miloucheva (Fraunhofer Institute), Ch. Brandauer (Salzburg Research), G. Panholzer (Salzburg Research) Expires: April 24, 2009 November 24, 2008 Policy based monitoring and learning of context for enhanced QoS guarantees draft-miloucheva-policy-monitoring-00.txt Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of 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 April 24, 2009. Abstract In order to enhance the policy provisioning and QoS guarantees, policy monitoring and context learning facilities are used. This document is focussed on facilities for monitoring and learning of context integrated in autonomous policy management system for enhanced user-centric Quality of Service (QoS) guarantees in heterogeneous Internet environment. The autonomous policy management includes components aimed at automation of QoS policy specification, provisioning and adaptation based on common policy repository. Integration of monitoring and context learning facilities allows to consider measurement and context data for more efficient policy provisioning and adaptation. The goal is to collect measurement and context information, in order Miloucheva Expires April 2009 [Page 1] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 to support automated policy decision and adaptation of actor's policies, when specific events (patterns) are detected in a given network context. Table of Contents 1. Introduction................................................ 2 2. Terminology used in this document........................... 3 3. Overview of QoS Monitoring and context learning facilities.. 4 3.1. Architectural considerations................................ 4 3.2. Policy monitoring .......................................... 5 3.3. Optimisation of QoS monitoring ............................. 6 3.4. Context learning ........................................... 7 4. Usage of Policy monitoring and context learning in QoS policy management scenarios ....................................... 8 4.1. NetQOS system for autonomous QoS policy management ......... 8 4.2. Integration of policy monitoring in NETQoS ................. 10 5. Further work ............................................... 11 6. References.................................................. 11 7. Author's Addresses.......................................... 13 8. Full Copyright Statement.................................... 13 9. Intellectual Property ...................................... 14 1. Introduction Facilities and tools for automated policy monitoring and learning of context are described, which are used for autonomous user centric QoS policy management in heterogeneous Internet environment based on dynamic policy specifications by different actors. The goal of the automated policy monitoring and context discovery is to facilitate the automated provisioning and adaptation of QoS policies of different actors considering performance and context events. Such events can be described by parameter values, thresholds and patterns. The background is the QoS policy framework for heterogeneous Internet environment (HQPIM) of the NETQOS architecture (see, [1], [2]). Its goal is to support QoS selection and mapping for different actors (users, operators, network providers) in heterogeneous networks including fixed, mobile and sensor technologies. The QoS policy framework for heterogeneous Internet is derived from the Policy Core Information Model (PCIM) [3], the Common Information Model (CIM)[4] and the QoS Policy Information Model(QPIM)[5]. The QoS policy monitoring and context learning facilities are based on: - Measurement and learning of context according to specifications of policy related performance parameters and patterns; - Collection, filtering and aggregation of policy related monitoring and context data in data bases; Miloucheva Expires April 2009 [Page 2] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 - Detection of policy violations and policy change requirements based on monitoring and detection of policy related parameters; - Interaction procedures and signalling between the QoS policy management system and the components for policy monitoring and context learning. This draft is focussed on the architectural considerations for QoS policy monitoring and context learning, as well as on the basic integration approach of the policy monitoring and context learning in the NETQOS autonomous QoS policy management system [1] [2]. The NETQOS system is based on business policy specifications and hierarchical policy refinement used to support preferences of different actors (users, operators) for QoS guarantees in heterogeneous Internet environment according Service Level Agreements (SLAs). 2. Terminology used in this document 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 [6]. This Draft considers the policy management terminology [7] and policy functions / components defined in NETQOS context [1]. Abbreviations used in the following text: MoMe - Monitoring and Measurement Manager CLM - Context Learning Manager MP - Measurement Policy MSC - Message Sequence Chart APM - Actor Preference Manager POLD - Policy Descriptor APA - Automated Policy Adaptor NetAgent - Component for automated configuration of mechanisms at network devices TransAgent - Component for policy configuration of mechanisms at transport entities. CM - Context Manager HQPIM - Heterogeneous networks QoS policy information model PCIM - Core Information Model QPIM - QoS Policy Information Model SLA - Service Level Agreement QoS - Quality of Service GUI - Graphical User Interface NETQOS - Policy management system for QoS guarantees in heterogeneous Internet environment Miloucheva Expires April 2009 [Page 3] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 3. Overview of QoS Monitoring and context learning facilities 3.1. Architectural considerations Policy monitoring and context learning is based on specification of business QoS policies by policy actors (users, network operators, service providers) and their mapping to intermediate and operational policies. Requirements for performance parameter monitoring and context learning are assigned to the business policy specifications of the policy actors and are mapped to concrete monitoring and context detection tasks. The basic architecture for QoS policy monitoring and context learning is shown in fig.1. +----------------------------------------------+ ! QoS policy management interface for ! ! Policy monitoring and context learning ! +----------------------------------------------+ ! ! ! v v ! +------------+ +-----------+ +-----------+ ! Measurement! ! Context ! !Signalling ! ! Monitoring ! ! learning ! !of policy ! ! manager ! ! manager ! !events ! +------------+ +-----------+ +-----------+ ! ! ! v v ! +----------------+ +-----------+ ! ! Monitoring ! ! Context ! ! ! Tools ! ! brokers ! ! +----------------+ +-----------+ ! ! ! ! v v ! +----------------------------------+ ! ! monitoring and context !------- ! data bases ! +----------------------------------+ Figure 1: Architecture for policy based QoS monitoring and context learning The architecture includes following components: - Measurement and Monitoring Manager; - Context learning manager; - Infrastructure for monitoring (passive and active monitoring tools); - Context brokers, context learning and identity management facilities; - Data bases of monitoring and context data. The interactions of the components allows the automated monitoring and context learning based on QoS policy specifications involving Miloucheva Expires April 2009 [Page 4] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 policy monitoring and measurement tools, as well as context brokers, identity management and other supporting facilities. 3.2. Policy monitoring Policy monitoring is aimed at detection of QoS policy violation for a given actor considering policy and SLA requirements. Policy monitoring actions can be assigned on different way to a given policy. One approach is to define one or more measurement policies related to given QoS policy definition [17]. A measurement policy specification (Pmeas) related to actor's QoS policy (Pqos) can be represented by: Pmeas (Pqos) = where - specifies constraints for the values of measured performance parameters in order to detect events, - defines corresponding measurement actions (scenarios); - specifies the procedures for event characterisation and signalling. An example for specification of a measurement policy is: ippm_loss > 0.2 OR ippm_1waydelay > 200 ms THEN call violation_event_loss_delay The measurement policy is used to detect violation of the requirements for performance parameter values, in this case delay and packet loss, related to the specified QoS policy of the actor and to adapt the operational policy by the QoS provisioning system based on the detected violations. Receiving the policy violation event, the QoS policy provisioning components can change the operational policies according to the SLAs and QoS policy requirements. Dependent on the application, network context and actor goals, different performance parameters can be considered for automated measurement and monitoring. The requirement for the policy monitoring and optimisation is to support flexibly appropriate sets of performance parameters, which can efficiently be assigned to the QoS policy of the actor. Parameters, which are considered currently in the NETQOS QOS policy monitoring, include: - One-way packet loss; - Maximum possible Throughput; - Round Trip Time; - Average Throughput; Miloucheva Expires April 2009 [Page 5] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 - Average link occupation on path; - Average router buffer occupation; - SrcIP/DstIP/SrcPort/DstPort/Transport Protocol of connections; - Sharing parts of the data path; - Path Bottleneck (congested router) identification. However, dependent on the context in heterogeneous Internet (mobile, sensor, broadcast) and application demands, further parameters can be considered for policy monitoring. 3.3. Optimisation of policy monitoring The optimisation of QoS monitoring is aimed at avoidance of redundant measurements. In order to avoid duplication of measurement and monitoring tasks, procedures for filtering and aggregation of policy based monitoring are required. The measurement aggregation and filtering procedures have to check dependencies of measured performance parameters and their related QoS policy specifications of the actors. Filtering and aggregation of measurements is based on analysis of following parameters: - The time interval, for which the QoS policy specification of the actor is valid; - Network context (source and destination for performing of the monitoring task); - Type of measurement and measured performance parameter; - Frequency of measurements; - Measurement actions; - Measurement conditions; - Time scope of the required measurement. When a new policy based measurement is required, then measurement policies (MP) are built and checked for redundancy (figure 2). +---------------------+ +------------+ +---------------------+ ! Monitoring data base! ! Filtering/ ! ! Monitoring data base! ! ! ! Aggregation! ! ! +---------------------+ ! of ! +---------------------+ ! MP(exec) !-->! measurement!-->! Updated MP(exec) ! ! MP(dep) ! ! policies ! ! Updated MP(dep) ! +---------------------+ +------------+ +---------------------+ executed Measurement Policies - MP(exec) dependent Measurement Policies - MP(dep) Figure 2: Optimisation of policy based measurements Reducing of redundancy of measurements is based on dependency analysis of measurement policies (MPs) related to the QoS specifications of the different actors, as well as adaptation of dependent MPs and actual executed MPs. The dependencies between MPs are updated each time, when a QoS policy with assigned MPs is entered in the system or changed. Miloucheva Expires April 2009 [Page 6] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 3.4. Context learning The QoS policies on business level are concise specifications of the QoS goals of the particular actors and are related to the SLA objectives. Context aware policy management deals with specification of user policies and their enforcement based on the consideration of context structure and behaviour [8]. Context dependent policy management is required in mobility scenarios [9], sensor [10], ambient networks [11], emergency [12] and other applications. The goal of the context learning is to automate the learning of the context based on QoS policy requirements and to signal the context changes to the QoS provisioning system in order to adapt the operational policy set according to the changed context. For instance, there are applications in mobile environment, which require adaptation of the QoS policy provisioning (e.g. operational policies) considering the actual location of the user of application. The context data is detected according to the QoS policy requirements of the user and is used for operational policy adaptation, when the context is changed. The context learning is to based on obtaining of knowledge in appropriate data bases about patterns and events related to the QoS policy content and SLAs. QoS policy related context can be defined based on: - User location; - User access network; - Characteristics of heterogeneous network devices (entities); - User end-systems and devices for the delivery of the service (application); - Time for service delivery; - Network events and their characteristics (anomay, failure, etc); - Patterns and models describing traffic, network QoS and routing behaviour. Dependent on the policy requirements, the context learning techniques can involve facilities for: - Detection of structural patterns and events concerning network traffic and other context data; - Detection of behaviour patterns characterising performance metrics; - Spatio-temporal routing pattern analysis [13]; - Detection of threshold and overload patterns [14]; - Modelling and prediction of context data (e.g. using of ARIMA and other models). The context learning is supported by context broker infrastructures and identity management facilities. The context learning is important component of the identity enabled QoS policy management architectures [15]. Miloucheva Expires April 2009 [Page 7] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 4. Integration of policy monitoring and context learning 4.1. NETQOS system for autonomous QoS policy management The NETQOS architecture is aimed at dynamic and autonomous QoS policy specification and provisioning for heterogeneous Internet environment [1],[2]. Policy abstraction and mapping from business level to intermediate, operational and configuration level policies is used in order to support the technology and vendor dependent configuration of QoS mechanisms. The NETQOS components for automated QoS specification and provisioning (figure 3) include: - Actor Preference Manager (APM) for actor and scenario oriented business level QoS policy specification; - Policy Description and Management (POLD) allowing unified policy access / storage; - Automated Policy Adaptor (APA) for policy decision, enforcement and adaptation; - Monitoring and Measurement (MoMe) for policy performance analysis and assessment of policy performance; - NetAgent and TransAgent for policy configuration at the managed entities, such as router or transport protocols, considering their particular capabilities; - Context Manager for control of interactions between different policy management components. Policy monitoring Management & context dependency of of enforced policies interactions +------+ |-------------| | MoMe |<--- | CM | +------+ |-------------| ! v +------+ +-------+ +-------------+ Policy +-------->| APM | --->| POLD |<-->| APA | adaptation, actor's +------+ +-------+ +-------------+ enforcement business ! ! ! QoS ! ! ! policies ! ! ! interface Storage v Access v v +---------------------+ +------+ +----------+ | policy repository | | Net | | Transport| | (unified policies) | |Agent | | Agent | | policy translation | +------+ +----------+ +---------------------+ Configuration of operational policies Figure 3: Interactions of NETQOS system components for dynamic QoS policy provisioning Miloucheva Expires April 2009 [Page 8] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 The QoS policies are designed according the SLAs considering the role, knowledge and expertise of the particular actors (operator, user, customer). At the business level, QoS policies are entered dynamically by the actors using APM GUIs. Ontology is used for QoS policy specification and translation of business QoS policies to intermediate, operational and configuration policies. The interoperation of the components is based on common policy repository for unified (intermediate) policy specifications. The access to the policy repository is performed by the POLD component. POLD functions translate and store the business policies of the actors as intermediate (unified) QoS policies in the repository. When the policy enforcement is required (by launching of the actor's policy controlled application), the APA (Automated Policy Adaptor) component obtains the intermediate policies and transforms them into operational policies (represented as XACML messages). APA includes functions for: - Mapping of intermediate to operational policies; - Adaptation of operation policies considering context and policy monitoring events; - Enforcement of operational policy configuration invoking the corresponding agents at transport and network level. The translation of unified QoS policies into operational policies considers SLAs and actor dependencies, as well as monitoring and network context. The operational policies are defined for managed entities and allow the mapping of the unified policy requirements to the specific QoS mechanisms of the managed heterogeneous entities (routers, services protocols). Operational policies are expressed by XACML (eXtensible Access Control Markup Language) [16] and applied for different kind of heterogeneous devices (routers and/or Transport level entities) taking into account the specific capabilities of the heterogeneous entities. Operational policies define QoS mechanisms assigned to user's applications, for instance guaranteed bandwidth allocation per application, QoS class assignment to user's flow, appropriate composition of transport functions for the application. Based on policy monitoring and context learning the content and assignment of operational policies to specific QoS policy can be changed. Miloucheva Expires April 2009 [Page 9] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 4.2. Integration of policy monitoring in NETQOS Automated policy monitoring and context discovery facilities are integrated in the NETQOS system architecture as Measurement and Monitoring (MoMe) component. The purpose of MoMe component is: - To evaluate the policy performance based on measurement policies, which are defined considering SLAs and QoS policy specification; - To signal the policy performance degradations, based on which the policy adaptation functions can adapt the parameters of the operational policies - To filter and aggregate measurement policies in order to avoid duplicated measurements. MoMe can detect specific events related to the policy (i.e. congestion, overloaded connection, etc.), and send signalling message to the APA components for adaptation of operational QoS policy mechanisms. After the operational policy for an application is enforced by the NETQOS system, the MoMe component receives a message to trigger appropriate measurement policy and actions. The Message Sequence Chart (MSC) describing the integration of policy monitoring functions in NETQOS is shown: NETQOS System Policy Monitoring (Context Manager) (MoMe) ! ! ! --------------------------------->! ! 1. QoS policy based ! Assignment and execution ! application is launched ! of measurement policies ! ! dependent on the context ! ! ! ! !<----------------------------------! 2. Policy violation ! Signalling the NETQOS components ! is detected ! (APA) about policy violation ! ! ! ! ! !---------------------------------->! Change of measurement ! 3. Requirements for change of ! policies for the given ! measurement policy ! context ! ! ! ! ! ! Figure 4: MSC describing the integration of policy monitoring in NETQOS policy management system Miloucheva Expires April 2009 [Page 10] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 The policy monitoring functions are invoked by a signal message, when a QoS based application is launched and its policies are enforced (Step 1). Through the CM, the MoMe is notified by the APA of the activation of a new operational policy for a given and identified connection. Based on the stored QoS policies for the application, the policy monitoring assigns appropriate measurement scenarios and actions for evaluation of QoS policy related performance parameters. In case that policy violation is detected by the policy monitoring (Step 2), the MoMe tool sends a signal to the NETQOS system (APA). This invokes the adaptation of the operational policy. After performing of measurement policies, the APA can in some cases request change of monitored performance parameters assigned to the QoS policy (Step 3). The MoMe is signalled in this case by APA to configure the new policy monitoring scenario (e.g. measurement policies). 5. Further Work Automated QoS monitoring and content discovery is used in the autonomous QoS policy management for more efficient QoS policy mapping, decision and adaptation based on monitoring and context data. The first prototype of policy measurement and context learning facilities is implemented and integrated in the NETQOS system. Further work is aimed to integrate more powerful context learning facilities considering routing, location, identity management and content specific information. The optimisation of measurements - aggregation and filtering is another focus aimed to reduce the measurement overhead in Internet infrastructures. 6. References [1] EU IST project, Policy Based Management of Heterogeneous Networks for Guaranteed QoS (NETQOS), www.ist-netqos.org. [2] I. Miloucheva, D. Wagner, Ch. Niephaus, ?User centric QoS policy management for heterogeneous Internet environment?, ICT-Mobile Summit, Stockholm, Sweden, June, 2008. [3] B. Moore, E. Elleson, J. Strassner, A. Westerinen, ?Policy Core Information Model-Version 1 Specification?, RFC 3060, Febr.2001. [4] B. Moore, Policy Core Information Model (PCIM) Extensions, RFC 3460, January 2003. [5] Y. Snir, Y. Ramberg, J. Strassner, R. Cohen, B. Moore, ?Policy Quality of Service Information Model?, RFC 3644, Nov. 2003. Miloucheva Expires April 2009 [Page 11] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 [6] S. Bradner, "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [7] A. Westerinen, J. Schnizlein, J. Strassner, M. Scherling, R. Quinn, S. Herzog, A. Huynh, M. Carlson, J. Perry, M. Waldbusser, ?Terminology for Policy-based Management?, IETF RFC 3198, November 2001. [8] US Patent 7072956 - Methods and systems for context-aware policy determination and enforcement, US Patent Issued on July 4, 2006. [9] P. Bellavista, A, Corradi, R. Montanari, C. Stefanelli, "A mobile computing middleware for location- and context-aware internet data services", ACM Transactions on Internet Technology (TOIT), Volume 6, Issue 4, November 2006. [10] S. Schwiderski-Grosche, "Context-Dependent Event Detection in Sensor Networks", Fast Abstract in 2nd Intl. Conf. on Distributed Event-Based Systems (DEBS'08), 2008. [11] H. Moungla, ?A Context-Aware, Policy-based Framework for Ambient Network", IEEE Workshop on Policies for Distributed Systems and Networks, 2008. [12] U.Hofmann, A. Veichtlbauer, I.Miloucheva, "Policy Aware Communication for Emergency Evacuation", International Journal of Computer Science and Applications, Technomathematics Research Foundation, 2008. [13] P.A. Guitierres, I.Miloucheva, ?Integrating Inter-domain Routing Analysis in novel management strategies for large scale IP networks?, Next Generation Tele-traffics and Wired/Wireless Advanced Networking (NEW2AN'04) conference, February 2004. [14] D. Hetzer, I. Miloucheva, ?Adaptable bandwidth planning for enhanced QoS support in user-centric broadband architectures?, in Proceedings of World Telecommunications Congress, Budapest, Hungary, May 2006. [15] I. Miloucheva, D.Wagner, Ch. Niephaus, D. Hetzer, "User-centric identity enabled QoS policy management for Next Generation Internet", International Review on Computers and Software (IRECOS) Journal, July 2008. [16] OASIS eXtensible Access Control Markup Language (XACML) TC, XACML 2.0 Specification Set, XACML 3.0 Work in Progress. Miloucheva Expires April 2009 [Page 12] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 [17] P. A. A. Gutierrez, D. Wagner, I. Miloucheva, Ch. Brandauer, U. Hofmann, ?Policy based QoS MONITORING, Automated Learning Strategies for Policy Enhancement?, International Conference on Wireless Information Networks and Systems (WINSYS'07), Special session on QoS Policy Management, Barcelona, Spain, July 2007 7. Author's Addresses Ilka Miloucheva Fraunhofer Institute, SATCOM FOKUS,Schloss Birlinghoven 53757 Sankt Augustin, Germany Fax: +49-2241-14-1050 Email: ilka.miloucheva@fokus.fraunhofer.de Christof Brandauer Salzburg Research, Austria Phone: 0043 662 2288447 Fax: 0043 662 2288441 Email: christof.brandauer@salzburgresearch.at Georg Panholzer Salzburg Research, Austria Phone: 0043 662 2288449 Fax: 0043 662 2288441 Email: georg.panholzer@salzburgresearch.at 8. Full Copyright Statement Copyright (C) The IETF Trust (2008). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Miloucheva Expires April 2009 [Page 13] INTERNET-DRAFT Monitoring and learning of context November 24, 2008 9. Intellectual Property The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Miloucheva Expires April 2009 [Page 14] INTERNET-DRAFT Monitoring and learning of context November 24, 2008