Worst-case Ethernet Network Latency for Shaped Sources Max Azarov, - - PowerPoint PPT Presentation

worst case ethernet network latency for shaped sources
SMART_READER_LITE
LIVE PREVIEW

Worst-case Ethernet Network Latency for Shaped Sources Max Azarov, - - PowerPoint PPT Presentation

Worst-case Ethernet Network Latency for Shaped Sources Max Azarov, Standard Microsystems (SMSC) Background Deterministic QoS guarantees dictate a need for an analytical evaluation of QoS parameters: worst-case latency, worst- case


slide-1
SLIDE 1

Worst-case Ethernet Network Latency for Shaped Sources

Max Azarov, Standard Microsystems (SMSC)

slide-2
SLIDE 2

Background

  • Deterministic QoS guarantees dictate a

need for an analytical evaluation of QoS parameters: worst-case latency, worst- case latency variation.

  • Shortage in the existing research of the

QoS performance of the Ethernet with well-behaved end-points

slide-3
SLIDE 3

Approach

1. Define a mathematical model

– define bandwidth shaping period – define shaped sources, abiding to a bandwidth reservation on a shaping period

2. Illustrate worst-case latency configuration 3. Provide analytical proof of the worst-case latency formula 4. If needed, complement model with more details, go to step 2.

slide-4
SLIDE 4

Current results

  • Provided proof of the worst-case latency formula

for a network of sources emitting same-sized packets

  • Provided worst-case configuration and latency

formula for the extended network model:

– varying packet sizes – lower-priority interfering traffic – partial load and arbitrary shaping period – Internal routing delay – varying number of ports on switches

slide-5
SLIDE 5

Worst-case latency for shaping period = 1ms, 5-port switches N – number of switches, L – network load in %

slide-6
SLIDE 6

In Progress…

  • Add network model with more details:

– higher-priority interfering traffic – two or more classes of traffic – low-latency and higher-latency – with different priorities and different shaping periods – add formula for latency variation

  • Use results as a basis for a QoS-enabled

traffic routing and policing (reservation enforcement) architecture