Service Level Agreements for VoIP Alan Clark CEO, Telchemy - PDF

Please download to get full document.

View again

of 23
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report
Category:

Study Guides, Notes, & Quizzes

Published:

Views: 12 | Pages: 23

Extension: PDF | Download: 0

Share
Related documents
Description
Service Level Agreements for VoIP Alan Clark CEO, Telchemy 1 Agenda VoIP SLAs What s typical Why typical isn t ok Approaches to SLA measurement What to measure The trust problem Final thoughts 2 What is
Transcript
Service Level Agreements for VoIP Alan Clark CEO, Telchemy 1 Agenda VoIP SLAs What s typical Why typical isn t ok Approaches to SLA measurement What to measure The trust problem Final thoughts 2 What is an SLA? Agreed set of performance parameters that: Are measurable Can be easily related to application performance In theory - if the SLA is met then the customer is happy How well does this work for Voice over IP? 3 Service Provider Perspective Contractual SLA Keep SLA as general as possible Easily measured metrics Measure at service demarcation point SLA monitoring But Want customer to have good experience overall but minimize level of contractual commitment to this 4 Customer Perspective Voice is a mission critical application SLA should report all issues that affect service quality (i.e. don t want service provider to claim they meet SLA when service unsatisfactory) Need service provider to make it work SLA helps to focus service provider on delivering agreed quality levels Highly reliable voice service more important than refunds 5 Actual (typical?) VoIP SLA Jitter 20mS Loss 0.1% Latency 100mS Availability 99.9% What does this mean in practice? 6 Jitter 200 Average jitter level (PPDV) = 4.5mS 150 Delay (ms) Time (Seconds) 7 Packet Loss - also time varying mS Avge Packet Loss Rate 40 Average packet loss rate = 2.1% Time (seconds) 8 Leads To Time Varying Call Quality Bandwidth (kbit/s) High jitter/ loss/ discard Voice Data MOS Time Degraded Service Quality (DSQ) Event = MOS less than X for more than N ms 9 Delay Echo problems Conversational problems 5 4 MOS Score 3 55dB Echo Return Loss 35dB Echo Return Loss Round trip delay (milliseconds) 10 Why does echo make a difference? Gateway IP Echo Canceller Echo Round trip delay - typically 50mS+ Additional delay introduced by VoIP makes existing echo problems more obvious 11 Customer expectations of service quality Listening quality Clarity, no distracting noise/ pops/ distortion Conversational quality No noticeable delay or echo Availability Always available, does not drop calls Signaling quality Low call setup delay, features work 12 A Better VoIP SLA? 99.9% of calls/intervals have MOS-LQ 3.9 MOS-CQ 3.8 Degraded Service Quality Events 0.1/ hour [DSQ =.] Latency 100mS Availability 99.9% Based on either reference or actual endpoint Transient quality problems Also reflected in MOS-CQ Availability of media AND Signaling path 13 Enterprise Scenario IP Phones IP Phones IP VPN PSTN Gateway IP Phones 14 Measuring at ISP Demarcation - active test Test call Active Test Functions 15 Active test for IP Service SLA Uses VoIP calls, to ensure packets are treated identically to real VoIP calls Use a Reference endpoint - I.e. a fixed configuration, known, virtual IP endpoint Test: Peak times - to understand quality under load conditions Off-peak times - to detect problems before they impact users 16 Measuring at user desktop - passive test RTCP XR SIP QoS Report Embedded Monitoring Function 17 Passive test for IP SLA Most effective for end-to-end measurement Embedded quality monitoring function in IP endpoint Can measure service quality, signaling reliability Collect data via RTCP XR or SIP QoS reports 18 Service Level Metrics - mine or yours Common problem with SLAs Service provider measures SLA and reports that they meet SLA Customer uses different tools and finds that service provider does not meet SLA Who is right? 19 Overcoming the somebody else s problem problem Need Common measurement methodology Ability to make the same measurements at the same time in the same way Otherwise Results will differ and fingers will point Solution? A common (trusted) measurement methodology A shared (trusted) measurement function that is accessible to both service provider and customer 20 A Trusted SLA Monitoring Function Enterprise Summary Stats Active test agent Summary Stats Service Provider Non-intrusive monitoring Edge router 21 SLA Monitoring Function Ideally - locate in edge router on customer premise Measurement data available to both service provider and customer Provides both Non-intrusive per-call monitoring of live traffic Active test agent for scheduled testing and troubleshooting Measures SLA in terms of Estimated call quality level (MOS, R) DSQ (Degraded Service Quality) events Loss, jitter, discard, delay 22 VoIP SLA Management VoIP SLAs What s typical Why typical isn t ok Approaches to SLA measurement What to measure The trust problem Final thoughts? VoIP does not have such distinct service boundaries as traditional telephony - some form of shared data model is an essential addition to an SLA for collaborative troubleshooting across network boundaries. 23
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks