The IETF is standardizing a new network management protocol called NETCONF and an associated new data modeling language called YANG. The NETCONF protocol follows a remote procedure call paradigm and views management data as structured XML documents. An extensible set of protocol operations can be used to retrieve, copy, edit, and delete the configuration of devices and to realize configuration change transactions over multiple devices. The YANG data modeling language provides a compact and semantically rich notation for the definition of data models for use with NETCONF.

 NETCONF implementations are already available from industry leading router vendors and suppliers of tools for embedded systems. There is a ready to use collection of open source tools to process YANG data models. This tutorial provides an introduction into the fundamental concepts of the NETCONF protocol and the YANG data modeling language. Throughout the tutorial, live examples will demonstrate the working of NETCONF and YANG. The tutorial presenter has been involved in the NETCONF and YANG design teams and has recently carried out interoperability tests between different NETCONF implementations.

 Characterized by their flexibility to be deployed and functional in on-demand?situations, combined with their capability to transport a wide spectrum of applications and resilience to dynamically 멻eal?around failed network elements, mobile ad hoc networks (MANETs) are gaining rapid momentum both in the commercial and military arenas. Illustrative examples in the commercial sector include the need for establishing communications in disaster areas and/or rural places where it becomes difficult to deploy fixed infrastructures.㺧 In the military sector, MANETs are becoming the basis for the future network-centric warfare (NCW) paradigm as exemplified by the Future Combat Systems (FCS) and Warfighter Information Network-Tactical (WIN-T) programs.㺧㺧 The success of MANETs is however critically tied to their capability of transporting a wide spectrum of applications with varying quality of service (QoS) requirements or service level agreements (SLAs), and providing continued/un-interrupted service (i.e., seamless recovery) despite failures in the underlying network.

㺧This tutorial addresses the complex topics of fault and performance management for MANETs. Via the use of novel fault and performance management techniques that can adapt to the dynamics of the underlying MANET, the tutorial will describe techniques to provide resilience and service assurances to a wide spectrum of applications with diverse QoS requirements.㺧 The principal functions of fault management, namely network monitoring and root cause analysis, will be discussed. Root cause analysis, in particular, is a function that becomes much more complex in MANETs ?as compared to wireline networks ?due to the dynamic and stochastic nature of these networks. We will examine the complexities of analyzing faults in a MANET environment. Another very important aspect of fault management, self-healing, will also be discussed. Given that MANETs are more prone to failures than wireline networks, it is imperative that MANETs provide the ability to automatically recover from different types of faults, whenever possible. We will look at descriptions of various fault scenarios, and the corresponding self-healing actions that can be performed to recover from these faults using a policy-driven framework.

㺧The discussion of performance management will describe the functions required to manage the performance of applications in a MANET. Aside from the collection and aggregation of performance statistics, which is a requirement for any type of network, a performance management system for MANETs must also be able to provide quality of service (QoS) guarantees for the applications running in the MANET. Unlike wireline networks that are typically characterized by an abundance of capacity, MANETs generally provide very limited bandwidth that needs to be carefully managed so that higher priority traffic receives better treatment than lower priority traffic when there is resource contention. To further exacerbate the problem, many MANETs have a mix of un-encrypted (red) and encrypted (black) network segments to capture varying security requirements along an application뭩 end-to-end path, resulting in very limited (if not no) visibility into the encrypted network segments.㻍 The QoS guarantees however have to be provided along an end-to-end path, despite the presence of any intermediate black network segments.㻍 Therefore, an important aspect of performance management in MANETs is the provision of service assurances to high-priority applications, sometimes at the expense of lower-priority applications, keeping in mind any security-related (i.e., red-black) restrictions. This section of the tutorial will provide background on this problem, including the challenges in providing end-to-end service quality in MANETs characterized by heterogeneous networking technologies and encrypted and un-encrypted network segments. An approach to providing quality of service using adaptive measurement-based admission control and dynamic quality adjustment will be discussed in detail, including the required functions and related algorithms that support management of QoS in MANETs.

㻍Feedback control is central to managing computing systems and networks. For example, feedback is employed to achieve response time objectives by taking resource actions such as adjusting scheduling priorities and bandwidth allocations. Unfortunately, software practitioners typically employ an ad hoc approach to the design of closed loop systems, often with undesirable results such as large oscillations or slow adaptation to changes in workloads.
㻍

㻍In other engineering disciplines (e.g., mechanical, electrical, and aeronautical engineering), control theory is used to analyze and design feedback loops. Control theory provides a way to determine if feedback loops are stable (e.g., avoid wild oscillations), accurate in their control (e.g., achieve the desired response time objectives), and settle quickly to their steady state values (e.g., to adjust to workload dynamics). Recently, control theory has been used in the design of many aspects of computing, with a few examples of commercial products designed using control theory. Examples of where control theory has been used include: networking protocols (e.g., new versions of TCP/IP), real time systems, web servers, database servers, multi-tier computing systems, and workload managers.

㻍This tutorial provides an introduction to control theory for researchers and practitioners with a background in computer science. The tutorial provides a short introduction to the basic elements of control theory, and then focuses on recent advances in both theory and application. The tutorial concludes with a discussion of research challenges.

1. Core concepts in control theory and its application.
2. Using control theory to optimize resource management and its application to self-tuning memory management in IBM뭩 DB2 Universal Database Management System.
3. Theory of predictive control.
4. Application of model-predictive control in distributed real time systems.
5. Scaling control solutions to virtualized data centers.
6. Managing power and performance in data centers with multiple feedback loops.
7. Research challenges.

㻍Mobile Ad hoc Networks (MANETs) are rapidly gaining in importance, both in the commercial as well as the military arenas. In order to deploy and maintain these networks effectively, it is imperative that appropriate network management techniques and tools be used. While a significant amount of research has been dedicated to the development of networking technologies for MANETS over the past decade, not much attention has been paid so far to the unique management needs of these networks. This is mostly due to the fact that since MANET technologies are relatively new, the bulk of the research efforts in the area of MANETs were concentrated on solving fundamental problems in MANET networking, such as routing, mobility management, transmission schemes, and so on. However, now that technologies for implementing and deploying MANETs are becoming more mature, it is time to turn our attention to the effective management of MANETs.

㻍Mobile ad hoc wireless networks differ fundamentally both in functionality and capability from their static wireline network counterparts due to a variety of reasons, including random node mobility, unpredictable network dynamics, fluctuating link quality, limited processing capabilities, power constraints, etc. All of these characteristics give rise to a need for dynamic changes both in the functioning and management of the underlying network. Due to node mobility, network links are dynamic and of unpredictable quality, resulting in intermittent connectivity. Finally, an underlying problem is the scarcity of wireless network bandwidth. Unlike today뭩 wireline networks, where bandwidth is plentiful and links are reliable, mobile ad hoc networks typically have very limited bandwidth and are relatively less reliable, due to environmental effects.

㻍A great deal of work has been done in the area of network management under the auspices of standards bodies such as the ITU-T (International Telecommunication Union-Telecommunication Standardization Sector) and the IETF (Internet Engineering Task Force). The developed standards have resulted in valuable architecture definitions, abstractions, protocols, and process models for managing different types of networks. However, one aspect that has not been adequately addressed is the automation of network management, and the integration of different management functions. The traditional network management functions include Fault, Configuration, Accounting, Performance, and Security (FCAPS) management. To see why these functions need to be performed in coordination with each other, consider the following illustrative examples. Statistics collected for fault and performance management can be processed and analyzed by various management applications, leading to diagnoses of network problems. In order to fix such problems, there may be a need for repairing or replacing faulty equipment (in the case of network faults); or it may be necessary to re-engineer the network to add more capacity to deal with severe network congestion problems; and so on. Such requirements need to be addressed manually, due to the need to physically install or repair equipment in the field. However, many performance and some fault problems can be handled by network reconfiguration. For example, if a network link is severely congested, it may be possible to alleviate the problem by sharing the traffic load with another under-utilized link, or by reducing the amount of management traffic, etc. This can be accomplished simply by appropriately reconfiguring the network. Today, this is done manually by experienced network operators who examine outputs from fault and performance management systems and decide how to reconfigure the network appropriately. This is the fundamental problem with network management today: there is too much human intervention required to run a network. In order to reduce the cost of network operations, it is critical that human intervention be minimized by creating a feedback loop between fault/performance monitoring systems and configuration systems, and by specifying policies that regulate how the system should be reconfigured in response to various network events.

㻍This tutorial discusses the management challenges associated with ad hoc networks, and provides an in-depth description of how policy-based network management can be used for increasing automation in the management and configuration of mobile ad hoc networks. It describes the required components of a network management solution for such networks, using a policy-based management framework.

㻍Starting by describing typical business cases, this session will explore how to use device management instrumentation features to improve visibility into the network. We will discuss monitoring standards, communication approaches between the device and the NMS applications, and vendor implementations by taking Cisco as an example. Features covered in details include MIBs, NetFlow/IPFIX, IP SLA, NBAR, etc.㻍 We provide also an introduction to service level management and discuss how techniques for performance monitoring can be applied to both validate and ensure service level objectives are being met.㻍The tutorial will include several practical examples from data, voice, and security domains that illustrate those techniques, including monitoring of specific service level objectives and capacity planning.

㻍ISO/IEC 20000 is an international standard for IT Service Management(ITSM). First published in 2005, it is experiencing increasing attention and acceptance by IT service providers, as well as developers of ITSM tools, products and systems. For IT service providers, being able to demonstrate ITSM capabilities through an ISO/IEC 20000 certificate is fast becoming a crucial factor for winning new contracts - or renewing old ones.

㻍Background: Many of the most pressing challenges faced by IT service providers ?increasing the availability and reliability of IT services, optimizing the responsiveness of IT support, improving the perception of IT by the business㻍 ?cannot be adequately addressed by network and systems management alone. Realizing that organizational aspects play a vital role in IT service management, many IT managers have turned to ITSM process frameworks like ITIL, Cobit or eTOM for guidance. Some of these ITSM frameworks are perceived as de-facto standards, but ISO/IEC 20000 is the first international standard for ITSM, offering IT service providers the opportunity to provide evidence of their IT service management capabilities through an internationally accepted certification. ISO/IEC 20000 shares some concepts and terminology with ITIL, but is not dependent on ITIL or any other ITSM framework. Where ITIL covers possible solutions to a very broad variety of possible ITSM-related issues in a general, non-binding form, ISO/IEC 20000 concentrates on defining concise, auditable requirements for addressing the 뱈ust-haves?of ITSM.

㺀This tutorial will provide an introduction to ISO/IEC 20000. It will explain the purpose and structure of ISO/IEC 20000, cover all sections of the standard and outline the most important requirements specified by it. At the end of the tutorial, attendees will understand the fundamentals of ISO/IEC 20000, its role in the context of IT Service Management, and the similarities and differences between ISO/IEC 20000 and ITIL. They will be familiar with the process framework of ISO/IEC 20000, its definition of an ITSM system, and have gained insights into how these concepts can be applied in an IT service provider organization.

The tutorial requires no specific prior knowledge, though a basic knowledge of ITIL will likely be advantageous when discussing some advanced concepts and the relationship between ISO/IEC 20000 and ITIL.

㺀㺀 Luca Deri

㺀In the past few years computers moved from a few of processors to strong multicore systems. IP network stacks and packet capture have taken limited advantage of this major processor change with the result that the gap between network and packet processing speed is increasing. This is because individual cores are slower than previous generation CPUs hence mono-core packet processing leads to poor performance. This tutorial introduces packet capture and describes the main issues that need to be taken into account in order to efficiently analyze packets. It highlights where multicore can be used and how existing applications need to be modified in order to take advantage of it.

㺀Heterogeneity and scalability are two of the key challenges in managing complex networks. Heterogeneity corresponds to management variables at different network layers and different types of networks. Examples of heterogeneous variables include topologies, channel conditions for wireless networks, physical failures, malicious attacks, and high-level user inputs. These variables exhibit complex dependencies at different spatial-temporal scales. The second challenge is the scalability. A managed network can have a large number of subnets/customers, and thus thousands of heterogeneous variables, resulting in a huge amount of management data. Hence, open issues are:

- How to characterize the dependencies among a large number of heterogeneous variables?
- How to quantify the scalability?
- What approaches can scale to a huge amount of management data?

㺀More and more automated and intelligent approaches are applied to network management to tackle scalability and heterogeneity. Machine learning provides a basis to those intelligent approaches but its potential is yet to be explored. It is thus timely to provide a systematic introduction and example-applications of machine learning to network-management. This tutorial would hope to serve such a role, to stimulate further interest in this interdisciplinary area, and to share lessons learned and potential future directions.

This tutorial would be offered for the first time, combining knowledge in both machine learning and network management in academia and industry