Вы находитесь на странице: 1из 18

Scheduling

Packet Scheduling

Flow: A stream of related IP packets that results from a single user activity and requires the same QoS (e.g., audio stream, video stream)

Flow classification: needed in datagram networks to decide which flow a packet belongs to

Packets from multiple flows compete for same outgoing link. Which packets should be given preference? How many packets should be transmitted from a flow? Simple solution: First come best served Complex solution: Provide QoS guarantees:

Using queuing, service and dropping policies to provide flows with their required QoS

Slide 2

Flow Classification
Identify individual "flows" based on packet header information

Packet Flows

Classifier

Packets

To distinguish flows, classifier uses: Source IP address, Destination IP address, Source port, Destination port
Slide 3

Scheduling Goals
Sharing bandwidth Fairness to competing flows Meeting bandwidth guarantees (max and min) Meeting loss guarantees (multiple level) Meeting delay guarantees (multiple level) Reducing delay variations

Slide 4

Queueing Basics
A queue consists of a scheduling discipline and a drop policy input

drop policy: what is dropped upon overflow

queued packets

scheduling discipline: what packet gets sent next

Slide 5

Queuing policies

Slide 6

First Come First Serve


Packets enqueued into a common buffer Server serves packet from front of queue No fair sharing of bandwidth No flow isolation No priority or QoS guarantee

Slide 7

Priority Queuing
Multiple queues with priority 0 to n-1 Priority 0 served first Priority i served only if 0 to i-1 empty Highest priority lowest delay/loss, highest bandwidth Possible starvation of lower class

Slide 8

Fair queuing
Goal: Isolation of flows. Round Robin Problem: Flows with large packets get more bandwidth.

Deficit Round Robin

Slide 9

Weighted Fair Queueing

Slide 10

Weighted Fair Queueing


Problem: We need to serve a whole packet at a time. Solution: 1. Determine what time a packet, p, would complete if we served flows by GPS. Call this the packets finish time, F(p). 2. Serve packets in the order of increasing finish time.

Slide 11

Generalized Processor Sharing (GPS)


The ideal max-min fair scheduling

Visit each non-empty queue in turn Serve infinite small portion from each GPS is not feasible; we can serve only packets

Slide 12

WFQ Problem 1
WFQ, Packets of the same lengths. weights

10 Mbps

The table gives a list of different input traffic rates (in Mbps) at the four input queues. Fill in the resultant output rates.

Slide 13

WFQ Problem 2
WFQ scheduler is used, all packets arrived at time 0. List the packet labels in order of transmission, and for each packet give the actual time it finishes transmission on the link.
6*0.25=1.5 (6-1.5)/0.5=9

The packets and their finish times are A1, F4, C6, G8, D11, B17, E21.
Slide 14

WFQ Problem 2 (a)


Show how the GPS schedule changes if a new packet H of length 3 arrives at time 9 for the middle queue. What is the WFQ schedule in this case? What happens if H arrives at time 15

H3

The packets and their finish times are A1, F4, C6, G8, D11, H14, B20, E24.
Slide 15

Class-Based Weighted Fair Queuing (CBWFQ)

Assign weights per class of traffic provide ensured bandwidth to each class of traffic. Drop tail on overflow

Slide 16

Low-Latency Queuing (LLQ)

Combination of Priority Queue with Class-Based Weighted Fair Queuing

Slide 17

Scheduling Disciplines
First come first serve (FCFS) Priority (PQ) Fair Queuing (FQ) (Round Robin (RR))

Deficit round robin (DRR)

Weighted fair queuing (WFQ) Class based queuing (CBQ)


Class-Based Weighted Fair Queuing (CBWFQ)

Low-Latency Queuing (LLQ)

Slide 18

Вам также может понравиться