1 / 12

Why Should I Listen?

Why Should I Listen?. Session Title: QoS & Network Management for Successful Enterprise VoIP Deployment. In other words: How do I roll out VoIP with assurances over network quality while containing network costs?. Raymond Russell, CTO, Corvil raymond.russell@corvil.com.

cutler
Download Presentation

Why Should I Listen?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Why Should I Listen? Session Title: QoS & Network Management for Successful Enterprise VoIP Deployment In other words: How do I roll out VoIP with assurances over network quality while containing network costs? Raymond Russell, CTO, Corvil raymond.russell@corvil.com

  2. What Quality? - how can I be assured about network performance? How Much Bandwidth? – how do I know if I’m over-spending for no return in quality? VoIP Users WAN Service Provider Corporate Headquarters What’s The Business Imperative? • VoIP rollouts promise a rich user experience with substantial OpEx savings • But: • If WAN bandwidth costs are too high, network savings will not materialize • If WAN bandwidth is not sufficient, network quality & user experience will be poor • CIOs need a deterministic way to control both network quality and network costs for VoIP…. • BEFORE committing to rollout

  3. Problem is Lack of Knowledge, Not Bandwidth Bandwidth wrongly blamed for latency problems in Network Infrastructure • For decades LAN and WAN network design has focused on providing appropriate bandwidth, and has addressed performance problems by adding more. Network managers must now adoptnew procedures and tools to correctly diagnose network performance problems. • Businesses that do not consider latency issues when deploying new global applications across the WAN — or IP voice on the LAN or WAN — will face poor performance, reduced reliability, disgruntled users and, potentially, the failure of major projects for application deployment. • Where the problem is incorrectly diagnosed as too little bandwidth, it will be discovered that adding more bandwidth won't resolve the issue, and misdirected expenditures will never be recovered. Enterprise Networking Predictions for 2005

  4. Key is Understanding QoS, Bandwidth & Traffic as One • Adequate bandwidthisessential for application performance • The tighter the QoS target, the more bandwidth may be required, and vice versa • Depends on how well the traffic multiplexes • QoS, Bandwidth and Traffic (QBT) for the network are inter-dependent • Full control requires an integrated approach to QBT • You need to know your Bandwidth Requirement as it relates to your QoS objectives for your different traffic Too Much Bandwidth: Unacceptable OpEx Costs ? What is my Bandwidth Requirement to meet my QoS objective Bandwidth Too Little Bandwidth: Unacceptable Quality IP Network Traffic over Time

  5. Let’s Get Technical • In IP networks, QoS is determined by three basic criteria: • End-to-end delay • Hop-by-hop delay variation (jitter) • Packet loss • The primary cause of jitter and packet loss is the behaviour and performance of router packet buffers

  6. Packet Delay, Loss & Jitter in Router Buffers Queuing causes Delay and Jitter • Human beings are highly sensitive to excessive delay • Echo cancellation systems are highly sensitive to jitter • Most codecs are highly sensitive to packet loss No room!!! 6 5 4 3 2 1 Output link Router queue Full queues cause Packet Loss

  7. Real-Time Traffic Bursts Buffer Behaviour Is A Millisecond Phenomenon • Application traffic is bursty at short timescales • These bursts cause instantaneous queue build-up • Queue build-ups can take tens and hundreds of milliseconds to clear causing degradation in user-perceived quality • The feature of traffic on which application quality depends is totally invisible to mean-rate measurement • It is not possible to manage what you cannot control, and you cannot control what you cannot measure Bandwidth What you see today • Peter Drucker • Management Guru Mean Traffic Rate E.g., 300 Kbps over a 5-minute period Time

  8. Setting QoS Thresholds to meet Service Level Objectives • Different Applications can tolerate different levels of packet loss and delay, which are expressed as statistical limits • e.g. VoIP => .1% packet loss and 10ms delay for 99.9% of packets • These are referred to as Service Level Objectives (SLO) • SLOs may or may not form the basis of a legal contract, but if they are not met the application will not operate correctly • Current technologies do not help the network manager to meet these targets for application traffic

  9. What you need to know: CORVIL BANDWIDTH (CB) E.g., the CB is 460 Kbps for this application to achieve no more than 250ms delay & 0.1% loss New Approach to Efficient Control of IP QoS How Does It Work? • Allows the user to specify the maximum loss and delay limits for any traffic and input these on the router • Measures the key statistical properties of real-time traffic at the millisecond level • Reports the minimum bandwidth required (in bits/sec) to meet the quality targets • Basis for Network Dimensioning, Monitoring, Troubleshooting, QoS Optimization, Capacity Planning, Just-in-Time Bandwidth Provisioning, New Application Roll-outs, Service Provider Policing Too Much Bandwidth Bandwidth Real-Time Traffic Bursts What you see today Too Little Bandwidth Mean Traffic Rate E.g., 300 Kbps over a 5-minute period Time

  10. Case Study: Enterprise VoIP Corvil “What-If” After Adding VoIP (No QoS on Router) • Corvil “what-if” capability was used to determine the impact of VoIP on the existing network prior to deployment • 5 simultaneous VoIP calls with Silence Suppression were added to current traffic with a delay target of 40ms • The network manager quickly established the Corvil Bandwidth as 3.9Mbps • Next he ran a “what-if” scenario on Corvil to determine the bandwidth saving from turning on QoS in the router • He quickly discovered that by placing Voice in the priority queue and all other traffic in the best-effort queue, the overall bandwidth requirement could be reduced to 2.37Mbps Corvil “What-If” After Putting Voice Traffic in Priority Queue

  11. Who is Corvil? • Software & hardware company based in Dublin, Ireland with sales offices in New York & London • Founded in 2000, now at 75 employees • Patents based on core mathematics expertise – key insight: “measure directly the entropy of packet traffic” • Cisco Systems is an investor & partner • World Economic Forum ‘Technology Pioneer 2004’ – “technology with the potential to have a substantial long-term impact on business and society in the future” • For more information visit www.corvil.com

More Related