Solution 5

4
Internet Services & Protocols – Solution 5: QoS – 1. Scheduling: Suppose that the Weighted Fair Queuing (WFQ) scheduling policy is applied to a buffer that supports three classes, and suppose that the weights are 0.5, 0.25, and 0.25 for the three classes. a) Suppose that each class has a large number of packets in the buffer. In what sequence might the three classes be served in order to achieve the WFQ weights? Explain your answer. (For round robin scheduling, a natural sequence is 123123123...) One possible sequence is 1 2 1 3 1 2 1 3 1 2 1 3 … Another possible sequence is 1 1 2 3 1 1 2 3 1 1 2 3 1 1 2 3 b) Suppose that class 1 and 2 have a large number of packets in the buffer, and there are no class 3 packets in the buffer. In what sequence might the three classes be served in to achieve the WFQ weights? 1 1 2 1 1 2 1 1 2 1 1 2 … 2. Traffic Shaping / Policing: Consider the token bucket policing mechanism that polices the average rate r and burst size b of a packet flow. We now want to police the peak rate, p, as well. Explain, how the output of this token bucket can be fed into a second one so that the two token buckets in series police the average rate, burst size, and peak rate. Be sure to give the bucket size and token generation rate for the second bucket. Average rate r = how many pkts can be sent per unit time in the long run Burst size b = max. number of pkts sent consecutively (with no intervening idle) Peak rate p = how many pkts can be sent per unit time in the short run For the second token bucket, generation rate r = p and bucket size b = 1. Fakultät Informatik Institut für Systemarchitektur, Lehrstuhl Rechnernetze

Transcript of Solution 5

Page 1: Solution 5

Internet Services & Protocols– Solution 5: QoS –

1. Scheduling:Suppose that the Weighted Fair Queuing (WFQ) scheduling policy is applied to a buffer that supports three classes, and suppose that the weights are 0.5, 0.25, and 0.25 for the three classes.

a) Suppose that each class has a large number of packets in the buffer. In what sequence might the three classes be served in order to achieve the WFQ weights? Explain your answer. (For round robin scheduling, a natural sequence is 123123123...)

One possible sequence is 1 2 1 3 1 2 1 3 1 2 1 3 …

Another possible sequence is 1 1 2 3 1 1 2 3 1 1 2 3 1 1 2 3 …

b) Suppose that class 1 and 2 have a large number of packets in the buffer, and there are no class 3 packets in the buffer. In what sequence might the three classes be served in to achieve the WFQ weights?

1 1 2 1 1 2 1 1 2 1 1 2 …

2. Traffic Shaping / Policing:Consider the token bucket policing mechanism that polices the average rate r and burst size b of a packet flow. We now want to police the peak rate, p, as well. Explain, how the output of this token bucket can be fed into a second one so that the two token buckets in series police the average rate, burst size, and peak rate. Be sure to give the bucket size and token generation rate for the second bucket.

Average rate r = how many pkts can be sent per unit time in the long runBurst size b = max. number of pkts sent consecutively (with no intervening idle)Peak rate p = how many pkts can be sent per unit time in the short run

For the second token bucket, generation rate r = p and bucket size b = 1.

Fakultät Informatik Institut für Systemarchitektur, Lehrstuhl Rechnernetze

Page 2: Solution 5

3. IntServ:

a) Briefly summarize IntServ's objectives and features.

• framework to provide individualized QoS guarantees to individual application ses-sions in the Internet

• Key features:

–Resource reservations

–Admission control

• Provided QoS models:

–Guaranteed Service: firm (mathematically provable) bounds on the queuing de-lays that a packet will experience in a router

–Controlled Load Service: specifies that a call will receive “a quality of service closely approximating the QoS that same flow would receive from an unloaded network element” [RFC 2211]

b) What are the main difficulties associated with the IntServ model and per-flow reser-vation of resources?

Scalability: Per-flow resource reservation implies the need for a router to process re-source reservations and maintain per-flow state for each flow passing through the router.

Flexibly service: The IntServ framework provides for a small number of specified service classes.

3. Resource Reservation Protocol (RSVP):

a) Briefly describe RSVP's purpose.

• Internet signaling protocol used to perform IntServ call setup, i. e. reserve the nec-essary network bandwidth for an application's data flow between sender and re-ceiver

• Can also been used to coordinate DiffServ functions across multiple networks

b) Explain the terms “soft state” and “receiver-oriented” and the achievements described by them.

• Soft state: installed state times out (and is removed) unless being periodically re-freshed; Advantages: no explicit state removal, no orphaned states as well as auto-matically restoring states after crashes, no need for reliable signaling messages

• Receiver-oriented: the receiver of a data flow initiates and maintains the resource reservation used for that flow; Advantages: Receiver is in control of the traffic she receives, thus allowing different receivers to receive and view a multimedia multi-cast at different resolutions.

c) Describe the path and reservation message in detail.

Page 3: Solution 5

• Path Message

–Sender to network signaling, send from sender to receiver

– Informs routers on the way and receiver about the presence of a sender and the flow specification.

–Contains: sender description, destination address, traffic specification, previous hop, refresh time

–Routers receiving a path message store the flow specification in a flow soft state

–Receiver determines appropriate resources to fulfill the QoS requirements and an-swers with a resource reservation message

• Reservation Message

–Receiver to network signaling answering a path message

–Reserves resources between sender(s) and receivers

–Contains Flow Descriptor

–Flow Spec: defines QoS to be used

– IntServ QoS model

–R-Spec: QoS parameters

–T-Spec: Traffic parameters

–Filter Spec: defines data packets that should use the QoS

–Reservation model: no filter, fixed filter, shared explicit filter

–Set of allowed senders

–RESV message instructs each router on the way to perform the necessary resource reservation by updating the flow soft state. In case of resource unavailability, the reservation request will be rejected (Reservation error)

• Both, PATH and RESV messages have to be resend periodically to hold all states up-to-date (“soft state”).

d) Explain how RSVP handles heterogeneous receivers, i. e. receivers with different net-work flow capacities.

Heterogeneous receivers are handled by encoding the data transmitted into several layers, e. g. a base layer and an enhancement layer. For example a base layer could encode a multimedia stream at a rate of 20 kbps, whereas the enhancement layer could have a rate of 100 kbps. In this manner receivers with 28.8 kbps access could receive the low-quality base-layer stream, and receivers with 128 kbps could receive both layers to construct a high-quality stream. Note, that the sender does not need to know the receiving rates of all the receivers. It only needs to know the maximum rate of all its receivers. The sender encodes the multimedia stream into multiple layers and sends all the layers up to the maximum rate into the multicast tree. In order to not excessively waste bandwidth in the network's links, the heterogeneous receivers

Page 4: Solution 5

must communicate to the network the rates they can handle. Thus, the routers on the way can merge the necessary reservations for distinct receivers.