COMP3250:

COMPUTER NETWORKS
ASSIGNMENT 1

Question 1: IP header
I would choose option 2 following the scenario in the question above. The most reliable option
is to avoid fragmentation, and this would be the case if option 1 was chosen. If option 1 is
chosen, this would mean that the medium in which R1 and R2 are transmitting information is
subject to loss of datagram after encapsulation. This potential loss is because the fragmented
packets are processed and could be switched when reassembling resulting in performance level
dropping. 

Thus, router R1 should opt for option 2, as fragmentation of the original datagram and then
encapsulating each fragment is the more seamless and logical answer. The IP router can break
the IP datagram into multiple series of fragmented datagrams, in which each portion carries
data from the original IP datagram. The bit sign in the flag field of the IP datagram is
represented as a fragment itself, where the bit sign is a different fragment offset, resulting in
more security when dealing with fragmentation. Hence, this is all required when receiving
fragmenting datagrams and reassembling the packet.

Question 2: Intra-Domain Routing

The distance vector protocol uses features which essentially makes itself loop-free. This means
that that the distance vector protocol outlines that any packets transmitted will not be rotated
repeatedly through the same routers for multiple times, which limits it from being an infinite
loop. Furthermore, the distance vector protocol gathers information from the routing table. The
features are:

 Hold-down timer. The feature hold down timer can be used when preventing any router
from learning new information regarding an unreachable route. If any router is aware of
information about a route being down, it will initiate a hold down timer from its end.
This hold time refuses route updates for a couple of minutes, approx. 180 seconds, until
the timer expires or after a route retracts its statement of being down, hence preventing
the formation of loops.
 Split horizon. A split horizon can help make a distance-vector protocol loop free as it
halts the advertisement of a route to an interface in which it can be received. Essentially,
it stops a route from routing back onto the interface and is used on conjunction with a
poison reverse feature. The ‘next-hop’ or successor router can be defined as the best
route, as it knows the routing update from its previous router, as it would not advertise
the route back to its previous router as it has gained route information.
 Route poisoning. The last feature can be defined as any router connected directly to an
immediate route, which detects the route is down or has failed. The way it advertises
this is through an infinite metric, hence poisoning the route, as mentioned through the
name. Any router which received the update of a poisoned route would be aware and
would not use the route in future.

Thus, the distance vector routing protocol is unaware of routers in other areas, hence having to
gather information from the routing table. Once information and updates are received with the
help of the features mentioned above, the DVRP will share this information with other routers in
the area and enforce loop-free routes.

Question 3: Intra-Domain Routing

The link-state routing algorithm avoids the problems of loops during broadcasting or better-
known terminology – flooding, by having each specific node keep a collection of link state
package messages in a database. The original starting point of each LSP includes the identity,
link information regarding its changed status, and a sequence number. When a router receives a
link-state packet, it will check it alongside the database to see if the packet is current and has
been received or is ‘old.’ In both cases, there is nothing further, however, if a packet arrives with
a sequence number that has never been accepted or seen, the packet will be retransmitted over
every other link, leaving out the arrival interface.

It is still possible for routing loops to exist in link-state routing, albeit for a short time, if for
example one router has received a link-state packet but the other subsequent router has not.
This results in an ephemeral loop, as the routers could potentially have an inconsistent view of
the network and the differing routes. As it is only for a short time, once the link-state packet has
reached all routers, the loop will vanish.

A scenario illustrating the creation and dissolution of an ephemeral routing loop is:

A configured network topology has implemented OSPF – a link-state routing protocol.

 Area 1(Backbone): router 6 (R6), router 1 (R1) and router 4 (R4).

 Area 2: router 1 (R1), router 2 (R2), and router (R3).
 Area 3: router 6 (R6), router 1 (R1), and router 4 (R4).

In this scenario, area 1 will be the inter-area route. Area 3 will have type 1 LSA. And the border
router will have type 3 LSA. The ARBs will only accept type 3 LSA from the backbone interface.

The information that will be updated will be sent from router 1 to router. It will follow the path
R1R6R4R5. This will result in the creation of loop because R1 is preferencing R6 as
opposed to R3 due to the full adjacency view. To stop the creation of the loop, a loopback is
needed in R3.

Question 4: Intra-Domain Routing

The selection of the best path using the least cost is sometimes not the best solution and cannot
apply to all scenarios in intra-domain routing. As mentioned above, the shortest path routing
refers to finding a path with the minimum distance or in this case, cost. However, in some
circumstances, this is not the case. 

An example is a need for communication services and the rapid expansion and dependencies of
such services. This is tied to mobile communication, where users can use this platform from
anywhere and at any time. The availability of this service and more specifically this path is
dependent on multiple links and the reliability of these links, not necessarily the shortest path to
create the link. Although the shorted path has a minimum cost in forwarding the data, the
dependability and functionality of the path depends on multiple parameters such as direct
traffic from source to destination.

This is defined as a real-world problem, which can be translated towards large networks.
Essentially, communication services cannot keep adding nodes to a growing routing table at
each node. Additionally, as more nodes are added to the routing table, the number of failing
links increase, topology properties change, and multiple parameters catalyse and trigger the re-
building the network. With the increase of more nodes, these events will occur more frequently.

Hence, although communication must be delivered at a quick manner, the shortest path is not
always the best solution, as the messages must be delivered correctly according to the routing
protocol which depends on the efficiency of the route establishment between nodes.

Question 5: CIDR-Route Aggregation 

The range of IP addresses mentioned in the question can be summarized using an inclusive
summary. This is done by a single summarised route, that can be as small a range of addresses
as possible, whilst including all routes and subnets shown and the probability of including
subnets that do not currently exist.
The addresses displayed are class B, and the range of class B is from 128-191, and the address
182 falls in this range. Whereas the subnet masks are class C, as it is 255.255.255.0, and this
class C allows 254 hosts per subnet.

When converting the addresses to binary, the first 17 bits are the same. (The first two octets are
fixed, and the first bit in the third octet remains fixed as well). These 17 bits represent the
network portion can be noted as /17.

182.68.1.0 = 10110110.01000100.00000001.00000000

182.68.2.0 = 10110110.01000100.00000010.00000000

182.68.3.0 = 10110110.01000100.00000011.00000000

182.68.119.0 = 10110110.01000100.01110111.00000000

182.68.120.0 = 10110110.01000100.01111000.00000000

So as noted above, 182.68 is fixed for all network, ranging from 1 to 120. Also, the third octet,
the fist bit always remains the same. The CIDR notation will hence be, 182.68.1.0/17.

Question 6: CIDR-Route Aggregation

As stated above, the two enterprise routers R1 and R2 are connected to the two networks
24.2.2.0/24 and 24.2.3.0/24. As stated above, the interface of R1 encounters a failure to the
network 24.2.2.0/4. The reason why R1 still announces and advertises the summary route
24.2.2.0/23 to router R3 as it identifies a child route 24.2.3.0/24 and we can assume that R3
received two routes to 24.2.2.0/24 from both R1 and R2. As the two routes have identical
metrics, R3 can choose R1 as its next hop for packets being sent to 24.2.2.0/24. The routing
anomalies that arise from this scenario is that upon packets arriving at R1, these packets will be
dropped as R1 does not have an identified route to 24.2.2.0/24. This anomaly is labelled as a
black hole, as route aggregation can result in packets being lost, regardless of existence and
valid path.

Routing anomalies are more likely to surface from this scenario as black holes for child routes.
This occurs per the creation of a sink route. When a router summarises an aggregate route with
blackhole, the packets will be sent to the child route and may arrive unexpectedly and as a
result, packets might get dropped due to sink routes. Packets sent to unallocated routes should
be removed and dropped immediately, however, if the child route in the scenario above is
allocated to a different router due to failure, a black hole is created.

Thus, the presence of a blackhole – incoming and outgoing packets are dropped without
acknowledgement that the packets did not reach its destination, and a sink route – routing
implementation automatically creates routes whilst advertising aggregated routes are difficult
to detect. This is due to the advertised aggregated route being selected over all other legitimate
paths which are allocated to a child route.

Question 7: Intra-Domain Routing-BGP 

The only way BGP can be configured to help the customer achieve their objectives is defining
what a BGP is, and the properties of a BGP. The routing table contains the following properties:
  Network IP address for the router or the host.
 Next Hop which identifies where to forward the data on the router.
 Path: Contains all the information in the path that has travelled until the current point in
time.
 Multi-exit discriminator which is for the external neighbour and the preferred path in
the network. A lower value of MED is preferred over a higher value.
 Local Preference: This is used to differentiate multiple paths, and the value with the
higher path is more preferred. This information is exchanged between all the internal
BGP routers.
 Weight: This property is similar to the local preference; however, it is only made
available and known to the router.

BGP can be configured to help the customer achieve the first objective by setting the next hop
for SYD router to N1 and for N2 to be set onto the MEL router so that all traffic will be
forwarded to N1 and N2.

For the second objective, the local preference attribute can be used. The higher the value of the
local preference, the higher the chances of the default route being taken. Hence, we set the
local preference attribute to MEL as it is the primary default and hence, it will be higher than
secondary default SYD. Ultimately, this will work, as MEL is selected as the primary default route
wherever there is a choice between the two.

For the third objective, for the customer to accept the inbound traffic via the primary route MEL
as opposed to the secondary SYD, the multi-exit discriminator would be used. This would set the
value of MEL at a lower point than the value of SYD, meaning that the SYD link will activate if
MEL experiences failure, which would be notified as the MEL would signify the lowest value.

Question 8: Multicast-IGMP
IGMP version 1 supports the query-response model, and there are two types of messages
available.

 Membership query: sent to 224.0.0.1 from the multicast router.

 Membership reply: sent to the multicast address of the group by the station which
wants to join the group channel.

The company has registered to received 8 different channels. When a host wants to join a
multicast group, it will send a membership message to a group address in which it wants to
receive. This results in s situation in which the multicast stream would feed towards a segment
after all stations have left that group. Hence, the above scenario, in which the active streams will
not have any mechanisms to find out if a group is active or not, as it is probable that all the
statins have left the group, and the active streams are still occupied by the group, as their
absence was not notified nor recognised by the router. The only way this will be recognised by
the router after a timeout value, indicating that there are no stations in the group, hence, it
would stop feeding the stream and bandwidth to that group.

Thus, the only way to overcome the problems is to implement IGMP v2 as opposed to v1, where
leaving messages are an indicator to overcome the above problems.