2020

1.

A.

I. (2p)

What is the relationship between the protocol data unit $P D U_{n + 1}$ and the protocol data unit $P D U_{n}$ for an upper layer $n + 1$ and a lower layer $n$

In transmission PDU n+1 will get wrapped by in a packet of the layer below. So PDU n will contain (part of) PDU n+1 in it's data section

When received the opposite will happen and PDU n will be unwrapped to get (part of) PDU n+1

II. (2p)

How does service data unit $S D U_{n}$ relate to the PDUs mentioned in the previous question?

The PDUs need to get passed down as SDUs to the service presented by the lower layer

The opposite will happen when received

B. (4p)

All routers on the left, all switches (networks) and point-to-point links on the right

C.

Follow the algorithms

2.

A.

I. (2p)

What is the purpose of an IPv4 converted address?

An IPv4 converted address is a way of converting an IPv4 to an IPv6 address, such that it can be carried over IPv6 networks. This way, one can decide to only run IPv6 in their network, while still allowing IPv4 hosts to use it (given you have the translation configured)

II. (2p)

What is the purpose of an IPv4 translatable address?

An IPv4 translatable address is meant for IPv4 hosts to be able to talk to IPv6 hosts, by translating the IPv6 address into the format of IPv4 through services like NAT64 or DNS64

Feature	IPv4-Converted	IPv4-Translated
Used by	IPv6 host	IPv4 host
Direction	IPv6 → IPv4	IPv4 → IPv6
Prefix	::ffff:0:0/96	64:ff9b::/96 (or similar NAT64 prefix)
Purpose	Represent IPv4 address in IPv6 format	Represent IPv6 address in IPv4 format
Used with	SIIT / NAT64	NAT64 / DNS64

III. (2p)

How are the addresses from questions 2(a)i and 2(a)ii collectively called?

IPv6-Emebedded addresses

B. (3p)

For the purpose of IPv4/IPv6 algorithmic translation a service provider has set aside the IPv6 prefix 2020:1218:fade::/48
What is the resulting IPv6 address (in canonical form) when embedding the IPv4 address 20.20.18.12?

20 -> 00010100 -> 14
20 -> 00010100 -> 14
18 -> 00010010 -> 12
12 -> 00001100 -> 0C

2020:1218:fade:1414:120c:0000:0000:0000

C. (3p)

Consider the subset of the IPv6 space given by the union of the prefixes

::/4
1000::/4
4000::/2
8000::/3
a000::/3
c000:/2
Describe this subnet as the union of prefixes with the maximum possible aggregation

0000
0001
01
100
100
10

So the first 2 combine into ::/3
The second one into 4000::/2
The last three into 8000::/1

D. (4p)

Mention four different functionalities offered by the Neighbour Discovery Protocol (NDP)

IPv6 to MAC translation
Router discovery
Redirects for route optimisation
Neighbour state (actively keeps track of reachability)

3.

A.

Not relevant

B.

Not relevant

C.

Not relevant

D.

Do STP

4.

A. (5p)

Not relevant

B. (2p)

What information about a certain destination network is distributed through the RIP routing protocol?

The target network prefix and the hop-count (metric)
(And next-hop in RIPv2)

C. (2p)

Which mechanisms are available to counter the slow "counting to infinity" problem?

Split horizon (don't re-advertise a route to the one you learnt it from)
Poisoned reverse (send a route with metric 16 if you know something to be unreachable)

D. (1p)

Which of the following timers is not defined in an RFC?

Flush timer
Hold-down timer
Invalid timer
Update timer

Hold-down timer

E. (2p)

What is the maximum value for the notion of infinity within IGRP?

255

F. (2p)

Which algorithm is used in EIGRP for loop free routing and fast convergence?

DUAL

G. (2p)

RIP version 2 uses, instead of broadcasting to 255.255.255.255:
___ to ___

multicast to 224.0.0.9

5.

A.

Dijkstra

B.

I. (1p)

Which routes are not carried in a stubby area?

No external routes are injected into a stubby area by the ABR
Instead, use a default gateway for all external routes

II. (1p)

What extra condition holds for a totally stubby area?

In a totally stubby area, not even inter-area routes are injected
One default route for internal AND external destinations

III. (2p)

Which multicast addresses does OSPFv3 use?

ff02::5 all OSPF speakers
ffo2::6: all designated routers

C.

I. (3p)

An LS request packet contains multiple occurences of three fields identifying a link state advertisement. What are those fields?

Type, link state ID and advertising router

II. (2p)

What are the two extra fields that make an LSA instance out of the link state advertisement in the previous question?

Sequence number and age

III. (3p)

What three flags are defined in a router LSA and what do they mean?

B: border router
E: external routing capability
V: virtual link endpoint

6.

A.

I. (4p)

AS 1:

Connected to AS 2 & 3
Three internal nodes, making for 3 iBGP connections
AS 2:
Connected to AS 1 & 3
Two internal nodes, making for 1 iBGP connection
AS 3:
Connected to AS 1 & 2
Two internal nodes, making for 1 iBGP connection

II. (3p)

If a certain AS has 20 eBGP peerings what can you say about the maximum and minimum number of iBGP peerings?
Suppose no use is made of the route reflectors or other optimisations

The number of eBGP peers says nothing about the number of iBGP peerings

If the AS has only 1 router, responsible for all 20 peerings, there will be 0 iBGP peerings
Alternatively, the AS can have any amount of internal routers, meaning an unbounded amount of iBGP peerings

B.

I. (2p)

The order of preference for using incoming routes is:

Customers before peers before providers

II. (2p)

What is the usual mechanism to make sure the order of preference is realised?

Set LOCAL_PREF accordingly (higher is better)

C.

What even

D. (2p)

With route reflectors, BGP information may need to travel over several iBGP peerings to reach every BGP speaker inside a given AS. What is the maximum number of peerings that might be needed in such a path?

That depends wholly on your network topology

If all iBGP peers are connected to one route reflector, then it is:
incoming -> reflector -> other peers
I.e. 2 peerings

But if you have many reflectors chained together it can be
incoming -> reflector -> reflector -> ... -> other peers
I.e. the number of reflectors + 1