SLA Monitoring by benbenzhou


									Mapping between QoS Parameters and Network
      Performance Metrics for SLA Monitoring

             Hyo-Jin Lee, Myung-Sup Kim and James W. Hong
              Distributed Processing & Network Management Lab.
              Dept. of Computer Science and Engineering
              POSTECH, Pohang Korea
              Email: {really97, mount, jwkhong}

             Gil-Haeng Lee
              SLA Management System Team
              Network Technology Lab.
              ETRI, Daejeon Korea
             Introduction
             SLM and SLA Monitoring
             QoS Parameter & Network Performance Metric
             Network Monitoring Methods
             QoS Parameter to NPM Mapping
             Design issue of SLA Monitoring System
             SLA Monitoring System Architecture
             SLA Monitoring for IP Backbone Network Service
             Conclusion & Future Work

2002 APNOMS                                 2                  DP&NM Lab
       Background
             Increased Importance of SLA
               • Liberalization and rapid evolution of the telecommunication market
               • Increased interest in service quality by customers
             Aggregated view of service behavior provided by SLA Monitoring
             Various research on network monitoring.

       Problem
             Lack of Systematic QoS Parameter to Network Performance Metric
             Lack of consideration Generalized SLA Monitoring System Architecture

       Goal of Our Research
             Define the concept of SLA Monitoring in the network layer
             Define QoS Parameter to Network Performance Metric Mapping
             Design of Generic SLA Monitoring System Architecture

2002 APNOMS                                          3                                DP&NM Lab
                                            Service Level Management (SLM)
                                                                                          SLA Creation
                                                                                               Create SLA template

                                                   Business Level Management              SLA Negotiation
            Customer Interface Management

                                                                                               Select applicable QoS Parameters
                                                   Service Level Management                    Negotiate the penalty in violation

                                                                                          SLA Provisioning
                                                SLA Creation      SLA Assessment

                                                                                               Configure the network element
                                                                                               Topology to provide the service
                                               SLA Negotiation     SLA Reporting

                                              SLA Provisioning   SLA Maintenance          SLA Monitoring
                                                                                               Verify the degree of SLA assurance
                                                        SLA Monitoring                         Surveillance of degradation or violation

                                                                                          SLA Maintenance
                                                  Network Level Management                     Analyze the reason of degradation.

                                                                                          SLA Reporting
                                                                                               Report the performance information

                                                                                          SLA Assessment
                                                                                               Customer Demand the payment and penalty

2002 APNOMS                                                                            4                                               DP&NM Lab
          SLA Monitoring

                    QoS Parameters
                     SLA Contract
                                                                      Problem Notification

                                                                         Performance Data

                                             SLA Monitoring

 SLA Monitoring is important                                             Network Monitoring System
 - Assure the degree of QoS Parameter
 - Use the monitoring result in reporting , billing and assessment.

2002 APNOMS                                                 5                                        DP&NM Lab
         QoS Parameters in Network Service

                                        Availability
                      Availability              The feasibility of service in every service request

                       Delivery         Delivery
                                                The percentage of packet loss or delay
 QoS Parameters                         Latency
                      Bandwidth                 The the round trip time from sender to target.

                                        Bandwidth
                        MTBF                    Used or available capacity

                        MTRS            Time based QoS Parameter
                                                MTBF ( Mean time between failure )
                                                MTRS ( Mean time to restore service )

2002 APNOMS                              6                                                DP&NM Lab
         Network Performance Metric (NPM)

              Availability                           Connectivity
                             Functionality                  The physical connection of network

                             One way loss            Functionality
                                                            Whether the associated network devices work
   NPMs                         RT loss                      well or not

                             One way delay           Loss
                Delay                                       The fraction of packet lost in transit from sender
                               RTT delay                     to target during a specific time interval

                             Delay variance          Delay
                                                            The time taken for a packet to travel from sender
                                                             to receiver
                               Bandwidth             Utilization
                                                            The throughput for the link expressed as a
                              Throughput                     percentage of the access rate

2002 APNOMS                                       7                                                 DP&NM Lab
         Network Monitoring Methods

       Monitoring                   Mechanism                            Related
        Method                                                           projects

         Active      By  setting up the test machine and sending     NIMI
                     extra traffic to another                         Surveyor
        Monitoring   System overload is very low                     NLANR AMP
                     Test packets can be lost in the low priority    PingER
                                                                      Skitter

        Passive      By capturing the packet                         CoralReef
       Monitoring     Analyze the captured packets                   WAND
                     Limited NPMs can be measured                    WebTrafMon
                                                                      NLANR PMA

          SNMP       Using existing SNMP agents                      Internet2
                     Only the throughput are measured                MAWI

2002 APNOMS                                    8                                   DP&NM Lab
         QoS Parameter to NPM Mapping (1)

               Measurement Mapping

              QoS Parameters                                        NPMs

               Availability                                     Availability

                Delivery                                                       One way loss
                                                                                 RT loss

                                                                               One way delay
               Bandwidth                                           Delay
                                                                                RTT Delay
                                                                               Delay variance
                 MTRS                                            Utilization

                 Evaluation Mapping

         Availability of                 device unfunctional time  disconnected time
         QoS Parameter         = 100                                                 100 (%)
                                                     total monitored time
2002 APNOMS                                                9                                    DP&NM Lab
          QoS Parameter to NPM Mapping(2)
 Q={q1, q2, q3, q4} : a set of QoS Parameters
 N={n1, n2, n3, n4, n5, n6} : a set of NPMs
                                                                          m(x)                      n(x)
 2N : the power set of N                                       Q                         2N                         E
 E={e1, e2, e3, e4, …} : a set of Evaluation                                                                       e1
                                                               q1                    {n1,n2}
 Functions                                                                                                         e2
                                                               q2                    {n2, n3}
                                                                                        .                          e3
                                                               q3                     {n4}
                                                                                        .                          e4
                                                                                     {n5,n6}                       e5
                                                                                        .                           .
                                                      QoS Parameter
                                                                                        NPM                Evaluation Function

 For each element in N,
 H, I, J and K should be considered
 m’: N         (H, I, J, K)                                             (N1,N2,N3,N4)

                                  H                       I                      J                          K
                                      Monitoring Method                                                        Agent Type
                                                              Monitoring Point       Monitoring Period
                                           (A,P,S)                                                           (aa, na, ta, sa)

2002 APNOMS                                                        10                                            DP&NM Lab
         Design Issues of SLA Monitoring System
             General Requirements of SLA Monitoring System
                 Should be designed with no dependencies on SLA provisioning.
                 Should be designed in a layered and distributed architecture
                 Should minimize the monitoring traffic and transferred data for
                  management purpose
                 Should provide extendibility and flexibility for the variation of the
                  number of customer and SLA contents

             Design Issues
                 Precise QoS Parameter to NPM Mapping
                 Decision of monitoring location, method, period and agent type
                 Decision of evaluation function for each Parameter
                 Design of communication method and data format between each
                  functional modules
                 Design of the stored data format
                 Design of the period to be kept in the storage.

2002 APNOMS                                         11                                DP&NM Lab
         SLA Monitoring System Architecture
    Related SLM Components
    Customer Business          SLA Maintenance         Reporting         Billing                     Service
         System                    System               System          System              Planning and Development

   SLA Monitoring System

                                   Problem              Data                                       Parameter
                                   Manager              Store                                       Mapper

        Customer                                     Performance
          SLA                                         Analyzer

                                       PM               PM                PM
                                     Collector        Collector         Collector

                            SLA          SLA            SLA             SLA          SLA
                           Agent        Agent          Agent           Agent        Agent
    Customer SLA                                                                                Parameter Mapper
       module                                    Monitoring module                                   module
2002 APNOMS                                                       12                                        DP&NM Lab
          SLA Monitoring for IP-Backbone Network Service
       Service Configuration

       Service Provider             Network Service Provider such as KT, AT&T
       Customer                     company, organization, other service provider, individuals
       Service                      IP backbone network service

       Measurement Mapping: m(x)
        QoS Parameter             NPM        method         point      period     type
       Availability        Functionality     active      edge router   5 min     na
                           Connectivity      active      edge router   5 min     ta
       Latency             RTT               active      edge router   5 min     ta
       Delivery            RT loss           active      edge router   5 min     ta

       Evaluation Mapping: n(x)
       QoS Parameter      Evaluation function                                                                IP-Backbone Network

       Availability               device unfunctional time  disconnected time
                          100                                                 100 (%)
                                              total monitored time
       Latency                           RTT                (m sec)                         Company     Company
                          total number of RTT test packet
       Delivery                      number of lost packet
                          100                                 100 (%)
                                  total number of test packet
2002 APNOMS                                                            13                                               DP&NM Lab
         SLA Monitoring for IP-Backbone Network Service

                                                      Parameter Mapper
    Customer Business                                 Map DB
    System                                            Customer SLA
    Service Planning and
    Development System                                Performance
                                                      Data store
    SLA Maintenance System                            Problem Manager
    Reporting System                                  Reporting Manager
    Billing System
                                                      PM Collector

                                                      Backbone router
                                                      SLA agent
                                                       - loss check
                                                       - RTT check
                                                       - connectivity check
                                                       - functionality check
                             backbone network

                                                     data flow

2002 APNOMS                                     14                   DP&NM Lab
         Conclusion & Future Work
        Conclusion
             Define the SLA Monitoring
             QoS Parameter to NPM mapping consists of
               • Measurement Mapping
               • Evaluation Mapping
             A parameter mapping guideline for the network service
             A generic SLA Monitoring system architecture and one

        Future work
             Concrete the evaluation function.
             Mapping guideline for other services
             Adaptation of SLA Monitoring system design for various

2002 APNOMS                                 15                         DP&NM Lab

To top