1. OverQoS: Offering QoS using Overlays Lakshmi Sahara Retreat, 2003

2. What is OverQoS? Embed QoS functionality in Internet via overlays –Overlay nodes implement QoS functions –No support needed from IP routers Challenges –Nodes not connected to congested points, –have no control over cross-traffic –cannot avoid losses (reducing sending rate doesn’t help!) Why Overlays? –Previous QoS architectures have yet to be deployed –Overlay-model empowers third-party providers to provide some form of QoS

3. How does OverQoS work? Entry Node Exit Node Overlay Traffic Redundant Traffic N-TCP pipe Step 1: Aggregate Loss and Bandwidth Control Flow 2 Flow 1 SchedulerRate Ctrl Step 2: Distribute bandwidth and loss amongst flows

4. How does OverQoS work? Overlay Node Overlay Node Overlay Node Flow 1 Flow 2Flow 3 Step 3: Provide QoS guarantees (bandwidth,loss) to a flow by “stitching” guarantees on overlay links. Step 4: How to perform QoS-routing of multiple flows with different requirements on an overlay network? Challenge: Links with varying bandwidth, loss Ongoing work!

5. Controlled-Loss Virtual Link (CLVL) Two parameters: –Statistical bound on loss rate, q (<= p; typically << p) –Capacity, c(t), possibly time-varying Can prove: if offered load < c(t), then loss rate < q How is c(t) determined? –Given f(t) to be the redundancy factor: c(t) = b(t)( 1- f(t)) Buffer mgmt & Scheduling & Traffic regulator Coder c(t), q De- coder b(t), p(t) Flow 1 Flow 2 Flow n OverQos Node control plane CLVL

6. OverQoS Services, Customers Without loss control With loss control 1.Spatial bandwidth redistribution 1.Reduced throughput with reduced loss 2. Temporal bandwidth redistribution 2.Statistical rate and loss assurances 3.Differentiated Services 3.Explicit congestion control OverQoS Customers: Streaming media server, Leasing overlay networks, Pipe abstraction for large institutions

7. Link char. FEC loss(%) FEC+ARQ loss(%) FEC overhead(%) FEC+ARQ (avgloss %) Loss Non- recoverable Cable (downlink) 0.050.0215.41.8 (1.8)0.0 Cable(uplin k) -best 0.030.026.30.44(0.4)0.0 Cable(upli nk)- worst 1.33.9951.333 (18.8)1.25 Umass- UCLA 3.82.8294.8(4.7)3.7 Umass- Gtech 0.010.07192.18(2.0)0.0 Ind-Berk 0.470.3201.0(1.0)0.42 Wireless (strong) 0.130.0741.6(1.6)0.0 Wireless (weak) 1.61.3224.26.2(6.0)1.5 UMass-UW 0.020.0328.51.8(1.8)0.0

8. Extractable Bandwidth

9. What bandwidth guarantees? Wide-area link b/w (1% guarantee) Aros(Utah)- Greece1100 Kbps Intel(Boston)- Lulea610 Kbps Korea - Greece630 Kbps Mazu(Boston)- NL510 Kbps Mazu – PDI(Bay Area)690 Kbps Cable Modem (down)490 Kbps PDI – Aros830 Kbps Lulea -Korea280 Kbps NL -Intel350 Kbps PDI –Lulea880 Kbps

10. Conclusions Inferences from our study: –FEC+ARQ has better overhead characteristics than FEC-based CLVL implementation –Loss recovery is not possible during large periods of bursts –OverQoS can be used to provide some form of statistical bandwidth guarantees Limitations: –Need more statistical measurements across overlay links –Many overlay links (except cable modems) are not lossy.