Chapter Preview
TopBackground
QoS is a well-understood and studied topic within the networking community. It is typically expressed in term of the following three metrics: the end-to-end delay, the jitter, and the packet loss rate. The end-to-end delay is the amount of time it takes to transfer a packet from the transmitter to the receiver, and it consists of a) the end-to-end propagation delay, b) delays induced by transmission systems and processing times inside the routers, and c) delays a packet encounters due to queueing in the buffers of the routers. Jitter refers to the variability of the inter-arrival times of the packets at the destination, and the packet loss rate is the percent of packets that are lost.
Different applications have different tolerance to these QoS metrics. Table 1 relates various common networking applications to the end-to-end delay and packet loss rate. For instance, for conversational voice and video it is important that packets should be delivered to the destination in less than 150 msec in order to maintain user satisfaction. (Studies have showed that in fact an end-to-end delay of up to 220 msec can be tolerated.) On the other hand, a packet loss rate of about 1 in 100 can be tolerated. That is, conversational voice and video type of applications are packet-loss tolerant but delay intolerant. On the other hand, a file transfer service is delay tolerant but packet-loss intolerant. This is because we do not expect a file to be delivered immediately, but the integrity of the file is important, and any lost packets have to be re-transmitted.
Table 1. QoS metrics for common networking services
Tolerance for
packet loss | Tolerant | Conversational voice and video | Voice mail | Streaming Audio and video | Fax |
| Intolerant | Remote app., command and control games | E-commerce
Web browsing | Texting,
File transfer
(foreground) | File transfer
(background),
email |
| | Interactive
delay<<1 s | Responsive
delay ~1 s | Timely
delay ~ 10 s | Background
delay >> 10 s |
| | Tolerance for delay |
Key Terms in this Chapter
Service Level Agreement (SLA): A contract between a network provider and a customer that spells out that spells out how much traffic per class the customer will submit, the customer’s expected QoS, and the price for the service.
Open Shortest Path First (OSPF): A protocol used to construct routes through the IP network that typically minimizes the number of hops (routers).
Quality of Service (QoS): It describes how well a flow of packets is served by the network. It is typically measured by different metrics, such as, end-to-end delay, jitter, packet loss rate, network availability, and bandwidth.
Call Admission Control: This is a scheme that assures that there is sufficient bandwidth in the network for a new flow of packets.
Software Defined Networks (SDN): A centralized solution for managing routers and Ethernet switches.
Explicit Route: A route through the IP network that satisfies a QoS criterion, such as, minimizations of the end-to-end delay, maximization of throughput, etc.
Voice Over IP (VoIP): A telephone service implemented over the Internet. Voice is packetized into IP packets which are transmitted to the destination over the Internet. At the destination, the voice data is recovered from the packets and played back.
Propagation Delay: This the time it takes for a bit of data to travel across the network from one node or endpoint to another.
RSVP-TE: A signaling protocol for setting connections in an MPLS network.
Enterprise Network: A privately owned and operated network.
DiffServ: A mechanism used in the IP network to provide QoS guarantees on aggregates of flows.