Presentation is loading. Please wait.

Presentation is loading. Please wait.

2 QoS and Network Management for Successful Enterprise VoIP Deployment Jeff Barker Packeteer Director, Product Management

Similar presentations


Presentation on theme: "2 QoS and Network Management for Successful Enterprise VoIP Deployment Jeff Barker Packeteer Director, Product Management"— Presentation transcript:

1

2 2 QoS and Network Management for Successful Enterprise VoIP Deployment Jeff Barker Packeteer Director, Product Management jbarker@packeteer.com

3 3 Experienced significant application performance degradation and misalignment Change in application performance problems in past 12 months Challenges Across the Distributed Enterprise 75% of IT Professionals do not know what applications are running on their networks Source: Network World / Packeteer Application Performance Survey -2003

4 4 Relative importance on using Quality of Service technology to deliver VoIP services. A. QoS is critical to delivering VoIP across the WAN B. QoS is necessary to support VoIP in the LAN and WAN C. Company places moderate importance on QoS though not critical to VoIP deployment

5 5  Your primary concern with VoIP is the unknown: – Data network reliability – Consistent voice level quality – Unknown impact to the data network  You want to deliver the same service quality as a traditional PBX/PSTN network  You expect reduced costs and complexity by consolidating voice and data on a single network infrastructure Key Issues You Shared With Us...

6 6 Critical VoIP Deployment Issues  Assessing Network Readiness for VoIP Deployment – Understand VoIP solution requirements – Assess Current Network Conditions  Current bandwidth utilization  Identify competing applications  Measure network and application performance  QoS Considerations for VoIP – The WAN bottleneck – The QoS Challenge – Migration to MPLS  Ensuring QoS for VoIP

7 7 Performance Considerations for VoIP  Packet Loss: no more than 1%  One-way Latency: no more than 150 ms  Jitter: no more than 30 ms  Bandwidth: 20 -100 Kbps per call CodecNominal Data Rate (kbps) Packetization Delay (ms) G.711641 G.729825 G.723.1m6.367.5 G.723.1a5.367.5

8 8 First find out what is there – Visibility  Visibility into applications – Identify and categorize the types of network traffic that will compete with VoIP traffic Physical Network Data Link Transport Session Presentation Application required classification traditional classification 1 7 6 5 4 3 2

9 9 VoIP Readiness – Measure Performance  Utilization/Throughput (Average & Peak) – By Link, Application, User  Application Performance – Response Time Network Delay, Server Delay, Latency (Round Trip Time) Worst performing clients & servers – Efficiency (Throughput-Packet Loss)/Throughput  Top Bandwidth Consumers – Applications (classes) – Users

10 10 Utilization Network Performance Monitoring Utilization  Utilization Metrics  Average bps  Peak bps  Bytes  Packets  Packet Sizes  Utilization Metrics  Average bps  Peak bps  Bytes  Packets  Packet Sizes NetWare 5 consumes a T1 for 2 days and impacts VoIP Performance Email bursts to consume whole link and impact VoIP & SAP performance

11 11 Response Times Application Performance Monitoring Response Times  Application Response Time Measurement – Total delay – Server delay – Network delay – Round-trip times  Distributions and Thresholds – See how delays distribute – Set thresholds to determine “good” performance – then track

12 12 Diagnostics Performance Monitoring Diagnostics  Diagnostic Metrics  Connection Initiations  Server-Refuses  Server-Ignores  User-Aborts  Unsolicited-ICMP  Diagnostic Metrics  Connection Initiations  Server-Refuses  Server-Ignores  User-Aborts  Unsolicited-ICMP Nimda MS Blaster starts up on int’l MPLS core link Quarantine the Virus Enables monitoring and alerting for virus & DoS attacks that are WAN based like MS Blaster, SQL Slammer, Nimda, Code Red and others.

13 13 WAN Access Link Bottleneck High-Speed LAN High-Speed Backbone 10/100/1000 56k - T1/E1 WAN QoS Challenges OC-3, OC-12, OC-48, OC-192 Propagation delay Queuing delay

14 14 A Typical Response – Add More Bandwidth  Temporary  Ineffective  Initial and recurring costs  May not be possible Applications will consume all available bandwidth Bandwidth will not go to critical applications Set-up and monthly expenses Higher speed links are not always available

15 15 The QoS Challenge Mission- Critical eMail File Transfers P2P Web Surfing SAP, Oracle, Citrix, etc... VoIP Real Jukebox, Doom, Quake, etc. TCP / IP Application- Neutral TCP / IP Application- Neutral + - - Time-Sensitive +

16 16 Migration to MPLS  Benefits – Central site connection does not handle intra-site traffic – decreasing latency – Fault tolerance – no single point of failure for VoIP calls I I I I I I Frame Relay MPLS

17 17 What Value Does MPLS Provide?  Ability to match application needs to cost of transport  Meshed networking with optimum routing  Convergence of voice and data on one network IP Telephony PoS Production control Email Web

18 18 Issues with MPLS  More applications than classes – hundreds of applications sometimes thousands – little knowledge of performance requirements susceptibility peak and average throughput acceptable delay parameters  Mapping is manually made – rate of change is increasing scalability? Premium Silver Platinum Best effort Gold MPLS Core Application to CoS mapping is not scalable

19 19 Issues with MPLS  More applications than classes  There are more applications on the network than anyone has bought  Non compliant classes, non compliant SLAs – MPLS components are not “flow aware” – even carriers that monitor, can’t control – SLAs don’t report on non-compliant traffic! Carriers can’t guarantee Application SLAs Voice traffic from CEO Voice traffic from customer P2P traffic on same port as voice Voice traffic from customer Premium CoS Excess traffic (frames) are randomly dropped into lower CoS

20 20 Issues with MPLS  More applications than classes  There are more applications on the network than anyone has bought  Non compliant classes, non compliant SLAs  Best effort can be good enough  Bottlenecks still remain CPE Router Bandwidth-Intensive Applications Latency Sensitive Business Applications Web Surfing Applications Peer-to-Peer Applications SP Router P2P, Unknown Oracle Notes Web Carrier MPLS Congestion MPLS does not help at the LAN-WAN bottleneck

21 21  Policies/Partitions – Protect critical traffic – Contain recreational traffic – Block malicious – Smooth disruptive bandwidth intensive traffic.  Strong Control – Set Per-Session Guarantees – Burstable configurations – Prioritized Access to Excess Bandwidth – Bi-directional control (inbound, too!!) Ensure VoIP Bandwidth at WAN Access Points

22 22 Enabling MPLS for VoIP – Monitoring, Marking, and Controlling  Ensuring VoIP Performance across MPLS Networks – Dynamically marking traffic for desired MPLS CoS – Eliminating delays from WAN access congestion – Monitoring application service levels 512Kbps 256Kbps BandwidthAllocation Router Best Effort Premium Assured Basic MPLS Label Tagging WAN Optimized MPLS Core Packeteer Router Cost Web VoIP SAP FTP

23 23  Planned deployment of large converged network with VoIP Oracle, SAP and other apps at 34 sites  Historically, ERP and CRM apps running slowly, bandwidth addition tried and failed.  Continue adding bandwidth?  Delay deployment of certain applications?  Deployed PacketShaper at all 34 sites on converged network  Deployed all apps at once, with quality assurance from PacketShaper  VoIP and all other apps running smoothly Victoria Dept. of Sustainability and Environment Situation Decisions Solution

24 24 ControlControl Packeteer System for Delivering Fast WAN, Fast Apps, Fast Business AccelerationAccelerationVisibilityVisibility  Determine source of problems  Determine which tools to employ to fix issues  Ensure availability of critical applications  Prioritize access to resources & provide quality of service  Accelerate performance of critical applications  Increase capacity & overcome limitations in protocol design  Classification of network traffic  Analysis of performance & utilization  Network diagnostics & troubleshooting  SLA audit  Application intelligent service class assignment (TOS/DS Marking)  End-to-end QOS & branch site fairness allocation  Availability management  Assured and accelerated performance of applications  Scalable network resources  Better ROI on IT

25


Download ppt "2 QoS and Network Management for Successful Enterprise VoIP Deployment Jeff Barker Packeteer Director, Product Management"

Similar presentations


Ads by Google