Lecture Outline TELE302 Network Design Lecture 3 - Quality of Service Design 1 Jeremiah Deng Information Science / Telecommunications Programme University of Otago July 15, 2013 2 Jeremiah Deng (Information Science) July 15, 2013 1 / 29 A QoS problem Jeremiah Deng (Information Science) July 15, 2013 2 / 29 Bag of Tricks Trick 1: Introduce priorities and classify the traffic. Routers use policies to treat packets of different classes accordingly. Problem remains: what if the phone application increases its data rate to 2 Mbps? Trick 2: Use marking and policing to ensure sources adhere to bandwidth requirements; to be done at the edges. Consider a phone application at 1Mbps and an FTP application sharing a 1.5 Mbps link. Will this cause a problem? Bursts of FTP can congest the router and cause audio packets to be dropped. What will be the solution? Jeremiah Deng (Information Science) July 15, 2013 4 / 29 Jeremiah Deng (Information Science) July 15, 2013 5 / 29
Bag of Tricks... cont. Bag of Tricks... cont. Trick 3: Isolate the flows from different applications / application types. Problem: can lead to inefficient use of allocated resources in certain flows. Problem: what if H3 now tries to make a call to H4, at 1Mbps? Trick 4: Employ admission control; disallow excessive traffic. Jeremiah Deng (Information Science) July 15, 2013 6 / 29 Jeremiah Deng (Information Science) July 15, 2013 7 / 29 QoS Phases QoS contract Client: specification of usage characteristics Service provider Service guarantee for that usage What if the client exceeds the contract? Discovery and selection Allocation Monitoring Jeremiah Deng (Information Science) July 15, 2013 8 / 29 Jeremiah Deng (Information Science) July 15, 2013 10 / 29
Network QoS Integrated Services Current Internet Best-effort service IETF proposals Integrated Service () Differentiated Services () Multiprotocol label switching (MPLS) RFC 1633, arguing against Bandwidth will be infinite. Simple priority is sufficient. Applications can adapt. A Three-step Model 1 Traffic characterization and specification RSpec - QoS requested TSpec - sender traffic 2 Signaling for call setup RSVP 3 Call admission per router Jeremiah Deng (Information Science) July 15, 2013 11 / 29 Jeremiah Deng (Information Science) July 15, 2013 12 / 29 Solving Congestion Resource Reservation Host based: Leaky Bucket Token Bucket Router Based: Load Shedding Choke Packets Resource Reservation Fair Queuing Primarily for connection-oriented networks During connection setup: Request resources from the network (bandwidth, buffer etc.) If the network has enough available resources to support the new connection, the connection will be established Otherwise, the connection will be rejected Once a connection is accepted, the host must use only the amount of resources reserved. It may not use more than that. Jeremiah Deng (Information Science) July 15, 2013 13 / 29 Jeremiah Deng (Information Science) July 15, 2013 14 / 29
Resource Reservation: Example Usage Policy Control Case 1: Source attempts to connect to destination, and attempts to reserve 4 Mbps for the connection Result: Connection accepted. There is enough bandwidth available. Available link bandwidths updated. Case 2: Source attempts to connect to destination, and attempts to reserve 5 Mbps for the connection Result:? What if the 4 Mbps host is malicious and attempts to go for 8 Mbps? Traffic Policing algorithm used to prevent non-conforming users from degrading network performance. Leaky Bucket smoothes out bursty traffic. Analogy: water is being poured into a leaky bucket which leaks at a constant rate. New packet is allowed if it does not overflow the bucket. Drain rate of bucket ensures policing. Jeremiah Deng (Information Science) July 15, 2013 15 / 29 Leaky Bucket Jeremiah Deng (Information Science) July 15, 2013 16 / 29 Leaky Bucket: Analogy Used in conjunction with resource reservation to police the host s reservation At the host-network interface, allow packets into the network at a constant rate Packets may be generated in a bursty manner, enter the network evenly spaced As a Traffic shaper : It changes the characteristics of packet stream. Packet stream more manageable and more predictable. Doesn t allow desirable bursty traffic. Jeremiah Deng (Information Science) July 15, 2013 17 / 29 Jeremiah Deng (Information Science) July 15, 2013 18 / 29
Token Bucket Token Bucket vs. Leaky Bucket The bucket holds tokens instead of packets Tokens are generated and placed into the token bucket at a constant rate When a packet arrives at the token bucket, it is transmitted if there is a token available. Otherwise it is buffered until a token becomes available. The token bucket has a fixed size, so when it becomes full, subsequently generated tokens are discarded Token bucket is more resilient to (short) traffic bursts. Jeremiah Deng (Information Science) July 15, 2013 19 / 29 Router Scheduling Jeremiah Deng (Information Science) July 15, 2013 20 / 29 Disadvantages of Integrated Services Inside each router Responsible to provide the reserved bandwidth Scheduling mechanisms Weighted fair queueing (WFQ): queue weights determine the percentage of bandwidth for each traffic class Priority queueing: transmit a packet from the highest priority class that has a non-empty queue Not scalable Routers process reservations per flow Not flexible Only a small number of pre-defined classes Host RSVP signaling Jeremiah Deng (Information Science) July 15, 2013 21 / 29 Jeremiah Deng (Information Science) July 15, 2013 22 / 29
Differentiated Services Differentiated Service Aggregation of flows in classes Scalability Two main functions Edge functions Packet classification Traffic conditioning Core function Forwarding Jeremiah Deng (Information Science) July 15, 2013 23 / 29 Differentiated Service Features Jeremiah Deng (Information Science) July 15, 2013 24 / 29 Traffic Classification and Conditioning Policing is the QoS component that limits incoming traffic flow to a configured bit rate Shaping is the QoS feature component that regulates outgoing traffic flow to a configured bit rate Marking packets DS field, e.g., TOS in IPv4 or Traffic Class in IPv6 Non-conformant packets Marked differently Shaped Dropped Jeremiah Deng (Information Science) July 15, 2013 25 / 29 Jeremiah Deng (Information Science) July 15, 2013 26 / 29
Differentiated Service Control Comparison of Congestion avoidance: drop policy Tail Drop Random Early Detection (RED) Weighted Random Early Detection (WRED) Congestion management: scheduling policy FIFO Fair Queuing Weighted Fair Queuing (WFQ) Class Based Weighted Fair Queuing (CBWFQ) Low Latency Queuing (LLQ) Jeremiah Deng (Information Science) July 15, 2013 27 / 29 Jeremiah Deng (Information Science) July 15, 2013 28 / 29 Recap QoS: principles, paradigms, and mechanisms References Oppenheimer, Top-down Network Design, 1 st Ed., Chapter 12 (Optimizing Your Network Design) TCP/IP Tutorial and Technical Overview, IBM RedBook, Chapter 8: Quality of Service Available from Course Schedule page. Next lecture: Network Simulation Jeremiah Deng (Information Science) July 15, 2013 29 / 29