Module 2.1: QA

1
Module 2.1 QA Putting it all together

• What is the max number of users/connections/sessio
  ns a particular network can support for handling
  a new connection?
• Delay Types
• Analyzing Delay
• Analyzing Throughput

2
Delay Types

• Propagation Delay This is the delay which is
  caused by distance. According to Einstein, whos
  views on the subject are still accepted today,
  communication across a distance cannot take place
  faster than the speed of light. Much of the
  communication we undertake today is carried by
  light. Electrical signals carried on wires also
  travel at approximately the speed of light. Fast
  as it is (300 million meters per second), the
  delay due to a pulse of light traveling from one
  side of the earth to the other is still
  significant. Around 15micro second. Usually, this
  is negligible, unless going through satellite
  links.
• Transmission Delay Transmission delay is much
  more significant than propagation delay over
  short distances. It is the delay caused by the
  fact that it takes time to feed a signal onto the
  communication line. For example, suppose the line
  is transmitting at 10 Mbit/s, and the user wishes
  to send a file of 10 Mbytes. 10 Mbytes is 80
  Mbits, so, all things being equal, the
  transmission will take 8 seconds. This is the
  transmission delay.
• Queueing Delay This, the most interesting delay
  is caused by storage and retransmission of bits,
  bytes, and packets, in equipment lying in the
  network between the origin and the destination of
  a transmission. Buffering can take place at the
  origin, and at any point along the way where
  retransmission takes place.
• Packetization/Processing Delay In packet
  networks, one more factor affects the end-to-end
  and round-trip delay for voice the delay
  inherent in storing up a whole packet of digital
  audio data in a packet.
• Others This may include Echo Cancellers (10ms),
  Codec for compressing speech (75ms for 2 ends).
  There also delays an the NIC and ports about
  15micro second (negligible).

3
Analyzing Delay

• Get idea of the new rate and packet size for one
  connection/user end-to-end.
• Show the Analytical model of two way traffic
  flow.
• In VoIP and Videoconferencing we assume the
  traffic in both ways is symmetric. Also, session
  can take place between two different parties
  (multimedia PCs). Point-to-point
  VoIP/Vieoconferencing.
• Calculated the delay for one session (VoIP or
  Videoconferencing). Plot delay for multiple
  sessions. As you add more sessions, network
  delay will start going up. The max sessions that
  the network can handle is when the delay goes
  beyond an acceptable delay value.

4
Analyzing Throughput

• The capacity of the network can also limit the
  number of connections/users it can handle for a
  particular type of service.
• This is determined by finding out the narrowest
  available bandwidth in the path. This is the
  network bottleneck.
• The narrowest bandwidth can be a router, switch,
  or link.

5
Network effective BW is the narrowest link
6
Network Capacity and Available Bandwidth
7
Example Multicast and Interactive Video
Streaming

• An videostreaming server is connected through a
  100Mbps Ethernet link to a switch.
• Switch processing is 2 million pps, and we assume
  it has unlimited number of ports. All links to
  the switch are full duplex. Link speed to all
  multimedia PCs is 10 Mbps.
• The video server has a video processing rate of
  200Mbps.
• The multimedia PC can process incoming and
  outgoing video packet to the application at rate
  of 25Mbps
• How many simultaneous sessions can such a network
  support, maintaining high QoS?

8
First
Multicast Video Streaming
9
Second
Interactive Video Streaming