Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

CS4105 Advanced Communication Networks and Systems > Lecture 2: 1st October 2019 > Flashcards

Lecture 2: 1st October 2019 Flashcards

Network QoS

1
Q

What does it mean for a network service to be best-effort?

A

Best effort refers to a network service that attempts to deliver messages to their intended destinations but which does not provide any special features that retransmit corrupted or lost packets. Thus, there are no guarantees regarding delivery. It does not provide any guarantee that data is delivered or that delivery meets any quality of service. An analogy can be made to the postal service.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the consequences of the simplicity of IP?

A

It is best-effort: can’t guarantee packets will be delivered, nor any QoS on the IP network (Internet).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is an autonomous system?

A

A network or a collection of networks that are all managed and supervised by a single entity or organization. They use a single routing mechanism to connect to the Internet.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is the Internet?

A

A global system of interconnected computer networks that use the Internet protocol suite (TCP/IP) to link end-user devices.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is a transit network?

A

A transit network is a network that bridges a connection between two other networks, as well as passes traffic for its own network.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is a transit AS?

A

Essentially an AS that acts as a transit network for two or more other ASes. An AS that provides connections through itself to other networks. That is, network A can use network B, the transit AS, to connect to network C. If one AS is an ISP for another, then it is considered a transit AS. An example would be a backbone provider.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is a distribution network?

A

Computer networks in the Internet which sit between the core network and access networks. They have a low level of multiplexing and more traffic than in access networks, towards the network edge, and less than the core network. They act to make content and services available to end users.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is the Internet’s core network?

A

Backbone infrastructure which acts to route IP packets across the Internet. It is composed of IXPs and NAPs, where ISPs and different geographic locations peer to each other on the Internet.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

How heterogeneous is the Internet?

A

Very

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is the purpose of routers?

A

To move inbound packets to the next hop, 1 step closer to their destination.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What should the scheduling algorithms in routers consider?

A

The priority of different flows and data types and the performance they require

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is FCFS?

A

First Come, First Served = simple scheduling algorithm that processes packets in a single queue and in chronological order.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What are the problems with FCFS in routers?

A

It fails to consider the lengths of packets and their purposes. This means it won’t wait for a large packet that has partially arrived before processing a smaller packet which started arriving later but has now fully arrived. This changes the delay between each packet in the flow, causing jitter.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What would a universal IP packet size fix in FCFS?

A

The jitter arising from having different sizes. A size too large would make packets slow to transmit individually and make it slower to transmit an amount of data < the max payload but would make it quicker to transmit more data and reduce the amount of traffic. The tradeoff is needless when you can just adapt for diff sizes.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is delay?

A

How long it takes for a bit of data to travel across the network from one communication endpoint to another.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is jitter?

A

Variation in the lengths of the delays between incoming packets in a flow. It indicates a disruption in the normal sequence of sending data packets.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What is data rate?

A

The amount of data sent over a flow per unit time. data rate, r = data sent / time = window size (sum of data size in each packet) / RTT

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What are some sources of delay?

A

propagation; end-system processing; transmission; buffering and queueing in rotuers; transmission

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What are some sources of jitter?

A

FCFS queueing; traffic aggregating; high load on routers; mixed traffic; dynamic changes in paths from sender to receiver

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

What is RTO?

A

RTO = retransmission timeout = the maximum amount of time to wait after sending a packet for its acknowledgment before resending the packet.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

How do you calculate RTO?

A

RTO = B * RTT est
RTT est = [A * last RTT est] + [measured RTT * (1 - A)]

B = 2
A = 0.9
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

What is loss?

A

When one or more packets of data travelling across a computer network fail to reach their destination.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

How do real-time and non-real-time applications respond to jitter?

A

Non-real-time applications can simply buffer and wait for packets to arrive to fill gaps but real-time applications will need to compensate for jitter, e.g. Skype skipping dropped frames.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

How do real-time and non-real-time applications respond to delay?

A

Both would just buffer and play at a delay.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

How do real-time and non-real-time applications respond to loss?

A

Non-real-time applications will just retransmit packets until they are delivered successfully. Real-time applications will use error correction and fill in missed data where possible.

26
Q

What are some sources of loss?

A

router congestion; network partitions or failures; time of day traffic; traffic from IRL events

27
Q

Why may packet reordering be interpreted as loss?

A

A packet with a higher sequence number would arrive before one with a smaller sequence number, which could lead to the receiver assuming the packet with the smaller sequence number was lost.

28
Q

What are some causes of changes in data rate?

A

changed network paths; changes in routing; congestion; high traffic

29
Q

What is a token bucket?

A

The token bucket is an algorithm used in packet switched computer networks and telecommunications networks which can be used to ensure that data transmissions, in the form of packets, conform to defined limits on bandwidth and burstiness (a measure of the unevenness or variations in the traffic flow).

30
Q

How do token buckets work?

A

The bucket fills at rate r to a maximum size of b, which it starts at. Packets given tokens by the bucket transmit <= the peak rate, p. The amount of data sent in a burst is less than (r * t) + b, where t = is the time taken for the burst to transmit.

31
Q

Why may it be useful to be able to reserve network resources on the Internet?

A

To ensure certain end-to-end performance criteria are met to ensure applications will work.

32
Q

What is RSVP?

A

RSVP = Resource Reservation Protocol = a signalling protocol within INTSERV that tries to reserve and describe resources across a network.

33
Q

What is INTSERV?

A

Integrated services is an architecture that specifies the elements to guarantee quality of service (QoS) on networks. INTSERV specifies a fine-grained QoS system, which is often contrasted with DIFFSERV’s coarse-grained control system.

34
Q

What does RSVP do?

A

Allows network users to reserve resources and for them and observers to have descriptions of information about the network.

35
Q

How is RSVP related to INTSERV?

A

RSVP is part of the INTSERV architecture.

36
Q

What is the difference between soft-state and hard-state?

A

Soft-state applications remove parts of state stored in nodes implicitly when they are not included in the next broadcast (after timing out). This response has to be triggered explicitly in hard protocols, whose states will otherwise remain unchanged forever.

37
Q

Is RSVP soft or hard state? Why?

A

Soft state as this is how network information is stored in RSVP

38
Q

What are the pros and cons of soft and hard state network protocols?

A

Soft state apps handle network failures or partitions better but leads to more traffic when refreshing states so they don’t time out.

39
Q

How does RSVP work?

A

The soft state representing the channel is gained by the sender sending path messages out, and the receiver sending resv messages. The resv messages give the sender the information needed to request network resources for the flow, which the receiver confirms are in place. The soft state and reserved resources can be updated with more messages.

40
Q

What are some problems with network reservations?

A

they require two passes, so one reservation which went just after another may be blocked despite it thinking it could be made; there is a lot of state data to hold and transmit; network failures would lead to a loss of reservations; you would need to modify all applications and routers; you would need to find a mechanism of billing users for reservations - this is v hard to do.

41
Q

Is RSVP used in networks today?

A

It’s used on some small, totally controlled networks, such as intranets, but not on the wider Internet.

42
Q

What is RSVP-TE?

A

RSVP-Traffic Engineering is designed to reserve network resources in a way that moves data between users in aggregated clusters

43
Q

How does DIFFSERV work?

A

Packets are classified according to a three-tier system, with QoS established between routers or network edges: packets are classified and marked by the network, not apps. Signalling, therefore, happens between routers rather than between hosts/apps.

44
Q

What is a SLA?

A

Service Level Agreement = a commitment between a service provider and a client that police the (performance of the) service provided and the method of charging.

45
Q

What is a SLS?

A

Service level specification = the part of a service level agreement (SLA), in which the standards required of a service provider are set out.

46
Q

What is a DS byte?

A

The differentiated services field in IP headers used instead of the outdated IPv4 TOS field, which set traffic precedence - priority. The DS field does the same thing/

47
Q

What are traffic classifiers?

A

Processes which categorise computer network traffic into a number of priority-based classes, by some system or set of measures.

48
Q

What are some types of traffic classifiers?

A

multi-field: DS byte + other header fields;
behaviour aggregate: DS byte only;
DS codepoint: values for DS byte

49
Q

What is expidied forwarding?

A

An RFC which extends DIFFSERV to give a VLL service, at data rates specified in the SLS. EF provides low loss, latency, and jitter.

50
Q

What is a VLL?

A

Virtual lease line = a way to provide Ethernet-based point to point communication over IP/MPLS networks.

51
Q

What is AF?

A

AF = Assured Forwarding = an RFC which classifies the probability of packets being dropped, and guarantees the customer that it won’t happen at all as long as within some limit of traffic.

52
Q

How does AF work?

A

Place packets into 4 classes, with further separation into 3 levels of precedence within each class. Classifies packets as to know the probability and precedence in which they will be dropped.

53
Q

What are traffic conditioners?

A

Traffic conditioners refer to classifiers, meters, markers, shapers, and droppers. They limit the bandwidth of connections.

54
Q

How does traffic conditioning work in DIFFSERV?

A

Policing conditioners discard traffic exceeding a certain rate and shaping conditioners buffers excess traffic.

55
Q

How does service invocation work in DIFFSERV?

A

Can be for subscriptions at the user/customer/site/group/organisation level or within organisations at the application/user/user-group level. IETF WIP on dynamically using RSVP from INTSERV.

56
Q

What are some problems with DIFFSERV?

A

no single standard could be agreed for SLAs –> different behaviour for the same packet classes –> defeats the whole purpose.

bottlenecks from the receiver may still limit the speed of applications even if you pay for a premium service, such as being a computer slow to process data you give them to reply with

57
Q

What is MPLS?

A

Multi-protocol label switching = a routing technique in telecommunications networks that directs data from one node to the next based on short path labels rather than long network addresses, thus avoiding complex lookups in a routing table and speeding traffic flows.

58
Q

What is segment routing?

A

The use of extra headers in packets that contain a list of segments: instructions that are executed on subsequent nodes in the network. These instructions may be forwarding instructions, such as an instruction to forward a packet to a specific destination or interface. Segment routing works on top of either an MPLS network or on an IPv6 network. In an MPLS network, segments are encoded as MPLS labels.

59
Q

Why is it almost impossible to deploy QoS mechanisms on the Internet?

A

Too difficult to get everybody to agree on how to provision network resources, classifications, billing, and it’s too difficult to get end-to-end guarantees across the Internet.

60
Q

What is an access network?

A

An access network is a user network that connects subscribers to a particular service provider and, through them, to other networks such as the Internet. It is at the edge of the Internet and further out than distribution networks. They have a low level of multiplexing and less traffic than in distribution networks.

61
Q

What is DIFFSERV?

A

Differentiated services is a computer networking architecture that specifies a simple and scalable mechanism for classifying and managing network traffic and providing quality of service (QoS) on modern IP networks.

62
Q

How do DIFFSERV and INTSERV compare?

A

INTSERV = integrated services; DIFFSERV = differentiated services.

DIFFSERV is not per-user, doesn’t use signalling, used tiered classifications rather than specifying network performance criteria, packets are marked by the network, not end-user applications, and reservations ar router-router or edge-edge rather than host-host (between endpoints).

CS4105 Advanced Communication Networks and Systems (18 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions