What is fragmentation?

The router does fragmentation if the packet (datagram) has a bigger size than the maximal transmission unit of the frame. Thereby the received datagram is divided into fragments.
How does the receiver know whether a packet is fragmented?
The fragments (except the last) will have the "more fragments" flag set. And the "fragment offset" field is nonzero (except the first will have 0).
How does the router on the receiver side reassemble the data from the fragments?
The router identifies the fragments with the same ID and reassembles the data through the fragment offset and "more fragments" flag.
Does the fragmentation occur on layer 2 (Data Link) or L3 (Network)?
Fragmentation occurs on L3. L3 must pass the data to L2 in a frame/packet size that matches the MTU of L2.
What is done when a host is about to send a datagram of 2'000 byte into a link that has a maximal transmission unit of 1'500 byte?
The router will fragment the datagram by the size of the MTU of L2.

What is the Count to Infinity Problem?

In Distance Vector Routing this problem is a routing loop that can occur when an interface goes down.
How does Link State Routing work?
Routers exchange messages to learn the entire network topology from which then the routing table is calculated with a shortest path algorithm.
What is the motivation for a routing hierarchy?
To reduce the size of routing tables.
What are the two levels of routing strategies?
Interior Gateway Protocol (IGP) and Exterior Gateway Protocol (EGP). IGP are protocols within routing domains whereas EGP are protocols between routing domains.
How can a device communicate over the Internet without an IP address?
This is not possible.
What is BGP?
The Border Gateway Protocol that makes routing decisions based on paths, network policies, or rule-sets configured by a network administrator.
How can we classify routing algorithms?
  • Distribution: centralized, decentralized or isolated
  • And a routing algoritm is either static or dynamic/adaptive.
IPv4: How do we know that an IP belongs to a class A network?
Class A addresses start with the leading bit 0 => address space from - Because if leading bits are '01', the address belongs to Class B. Note that the binary number 010000000=27=128010000000=2^7=128.
In IPv4, how can we connect more than 2322^{32} devices to the Internet?
Through the use of local addresses and NAT.
What is a multicast address?
A multicast is like a broadcast, but whereby only a group of hosts is targeted. In IPv4, multicast addresses start with 1110 ( -
What is the subnet mask of a class A network?
IPv4: How many hosts does class A, B and C allow?
A: 2242^{24}, B: 2162^{16}, C: 282^{8} (minus 2 as one address is the gateway address and the other the broadcast address of the network)

What is Classless Inter-Domain Routing (CIDR)?

CIDR is used for routing and allocates IP address more flexible.
What are the problems with Classful IP Addressing?
Either wasting alot of hosts or having not enough as there are only 3 main classes.
What is the goal of CIDR?
To slow the exhaustion of IPv4 addresses and to slow the growth of routing tables across the Internet.
How does CIDR slow the exhaustion of IPv4?
Less unused IP addresses in a CLASS A/B/C network through variable-length subnet masking. Possible through extensive use of Network Address Translation (NAT)
How can the following IP configuration: and be written with CIDR notation? / 29
How would the default subnet mask of a class C network look if we want to divide the address range into two subnets? => therefore we get two new subnets with 126 - 2 hosts each.

Where are local IP addresses used?

Inside a LAN, which is a private network. The local IP never appears inside packets outside of the LAN as local IP addresses cannot be routed through the public Internet.
Why were private address spaces originally defined?
Because of the IPv4 address exhaustion.
What is the difference between the loop-back address and localhost?
The same. The loopback address is also referred to as localhost.
In IPv4, how can we connect more than 2322^{32} devices to the Internet?
Through the use of local addresses and NAT.
Does IPv4 have any error control mechanisms?
IPv4 can only detect errors through the checksum that is part of every packet's header. If the checksum is invalid, the packet is discarded and a message is sent over ICMP.

What is IPv6?

IPv6 is the actual version of the Internet Protocol that was developed by the IETF. IPv6 is a draft standard since 1998 and became an Internet standard in 2017.
What were the design goals of IPv6?
more addresses, IPSec (security by design), simpler headers and better support for mobile devices
What are the reasons that IPv6 was not faster adopted?
  • NAT
  • Older telcos have still enough IPv4 addesses
  • IPv6 implementations have/had bugs.
What is IPv6 over IPv4 (6rd / 6in4)?
The transistion mechanism to migrate from IPv4 to Ipv4 encapsulates IPv6 traffic over IPv4 links. Uses stateless address mappings between IPv4 and IPv6 addresses.
What is IPv4 over IPv6 (4rd)?
Once the IPv4 to IPv6 transition is almost completed an IPv4 over IPv6 tunnel is used to connect deprecated IPv4 sites.
What is special about the ::1/128 IPv6 address?
What is the max number of IPv6 addresses?
The length of IPv6 addreses is 128 bits. Therefore, this allows 21282^{128} addresses.

What is a routing table?

A routing table contains rules that are used to determine where data packets over a network are directed.
What is the default gateway?
The node in a network that acts as a router. Nodes in the network send all packets to the default gateway when no other route matches the destination IP address of a packet.
What can you infer when the gateway of a destination address is in the routing table?
The gateway means 'unspecified'. Therefore, we know that the host (destination) is locally connected and no hop is needed.
In a routing table: What does the flag 'UG' mean?
U => the route is up and G => use gateway