Week 4 - Network Layer Flashcards

1
Q

What is routing?

A

the propagation of connectivity informaiton in order to build routing tables. note routing is never done upon packet receipt, but rather the routing tables know the paths apriori. Equivalent to people who put traffic signs on roads.

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

What is the routing plane/forwarding plane?

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

What are intra/inter domain protocols?

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

How does Dijkstra’s best path algo work?

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

What’s the difference between distance vector algorithms and link-state algorithms?

A

Distance vector (Bellman Ford Equation):
1. Distribution of aggregate information
2. Distribution only to neighbours
3. Example algos: RIP Protocol

Link-State (Dijkstras equation):
1. Distibution of informaiton on local links (every router knows about every link in the network)
2. Distribution to every node
3. Example algos: OSPF protocol

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

What is RIP?

A

Routing Information Protocol
- Application layer protocol running over UDP

How it works?
- A router running RIP broadcasts a routing message every 30 seconds.
- Each update contains a pair <network, distance> (where distance usually number of hops)
- All routers send equivalent messages and update if shorter path (note no change is made if hte distance is the same)
- If a router does not receive an update after 18- seconds the route is deleted.
- Max distance for RIP is 16

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

What issue arises with the RIP protocol and how is it resolved?

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

What is the OSPF protocol?

A

Open Shortest Path First = Flooding + Dijkstra

  1. Router maintains descriptions of state of local links as a directed graph
  2. Transmits updated state information to all routers it knows about using flooding (sends a message about each link to all neighbours, they replicate that on avoiding loops).
  3. Router receiving update must acknowledge (lots of traffic generated)
  4. Each router then calculates what is the shortest path to reach all the destinations, in turn calculating which link to be used for every network.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What are link costs?

A

The cost of each hop in each direction (routing metric)
OSPF allows for control over metric schemes, such as Minize monetary cost (tos 2), maximise reliability (TOS4).

Note:
Once a routing table is created, it can be simplified as we can say (for a given input, shortest path to a given output is always on this link, so the table can be optimised (flattened)).

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

What is an area?

A

Areas make large internets more manageable. Configured as a backbone + multiple areas.
Area - collections of contiguous networks and hosts + routers connected to any included network.
- Each area runs a separate copy of link state algo
- Topological databse + graph of only that area + broadcast of links
- Reduces traffic
- Intra area routing relies solely on local link state info
Backbone - contiguous collection of networks not contained in any area, their attached routers and routers belonging to multiple areas.

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

What are OSPF packet types?

A

Hello: used in neighbour discovery
Database description: defines set of link state information present in each routers database
Link state request
Link state update
Link state acknowledgement

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

What is IS-IS?

A

Similar to OSPF except:
1. IS-IS does not support point to multipoint links
2. IS-IS does not support virtual links
3. IS-IS outer can belong to only one area

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

How is OSPF Weight calculated?

A

There is no standard, usually use 10^8/(interface bandwidth)

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

What is BGP?

A

Border Gateway Protocol (BGP). OSPF/RIP discuss transfer of packets in routes inside an Autonomous System (AS). AS can be a big company (e.g. IBM), or an ISP etc.. BGP is how data gets transmitted between AS.

Messages are sent over TCP.
Three functional procedures:
1. Neighbour Acquisition
2. Neighbour Reachability
3. Network Reachability

BGP Messages
1. Open - start a neighbour relationship with another router
2. Update - Transmit information about single route, or list multiple routes to be withdrawn
3. Keepalive - acknowledge open messages, or periodically confirm neighbour relationship
4. Notification - send when error condition detected.

When we talk about BGP we usually talk about eBGP. iBGP is the part of BGP concerned with transmitting external connectivity information INSIDE the AS (note its not a routing protocol).

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

How do BGP and CIDR operate together?

A

CIDR allows BGP to transmit much less information due to aggregation of routes.

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

What are BGP Route Reflectors?

A

Some AS have a lot of external connections. If every router transmits every message theres a lot of traffic. So iBGP uses Route Reflectors - these receive messages and send summaries to all the routers.

17
Q

What are the most used BGP attributes? And how are they used for Route selection?

A

ORIGIN. tells how a route was learned (IGP, iBGP etc.)
ASPATH - list of AS numbers htat packets will traverse for this announcement
NEXT-HOP - IP address of the router that packets need to be sent
MULTI-EXIT DESCRIMINATOR (MED) - When an AS has several options to a given destination, it may include information for preferring one of them
LOCAL-PREF: Allows a specific AS to signal internal preference for a route to a destination (when there is more than one).

18
Q

What is a transit vs non transit service?

A
19
Q

What are the BGP routing policies?

A

customer-provider peering: customer C buys internet connectivity from provider P (transit)
shares-cost peering: domains x and y agree to exchange packets by using a direct link or through an interconnection point (non-transit)

The routing policy of a domain is implemented via the route filtering mechanism on BGP routers:
Inbound filtering - upon reception of a route from a peer, a BGP router decides whether the route is acceptable, and if so whether to change its attributes.
Outbound filtering - before sending its best route towards a destination, a BGP router decides which peers should receive this route and whether to changes its attributes before sending it.

20
Q

What can cause route instabilities in BGP and what methods are used to alleviate it?

A

Solutions:
Route Flap Dampening:
- if the link AS1-AS2 goes up and down frequently it will disturb the entire internet. In this case AS2 stops advertising routes to AS1 to the rest of the internet.
Redundancy with Multi-homing:
- Multi-homing refers to a single network having more than one connection to the internet (load-balancing, fault-tolerance). Two types: Single provider, multiple providers.
- When a customer has two providers (multi-providers). This has some side effects: destroys route symmetry, may cause packet reordering, makes firewalls difficult, CIDR becomes difficult.
- Load balancing (single-provider) - can be achieved by announcing some routes in one link and some i others. This can be done with Route Reflector configured to announce some routes one way and some the other.
Confederations:
- An AS broken into sub-ASes
- These can run different IGPs