Final

Sample Final

1. Overlay multicast. (20 pt)

(a) Name two advantages of overlay multicast over IP multicast.

(b) What are the targeted applications in the case of the end-system
multicast? But in the case of Overcast?

2. Active Networks. (20 pt)

(a) What is the main problem active networks try to address?

(b) Compare in terms of advantages and disadvantages active networks
and overlay networks. Name at least one advantage for each of them.

3. Internet measurements. (20 pt)

(a) Give two reasons of why the default path selection in today's
Internet is not optimal.

Sample Final Solutions

1. Overlay multicast. (20 pt)

(a) Name two advantages of overlay multicast over IP multicast.

(b) What are the targeted applications in the case of the end-system
multicast? But in the case of Overcast?


Answer:

(a) - It does not require infrastructure changes; It can be deployed
      in today's Internet without any changes. 

    - It is possible to implement application specific
      functionalities. See below (b)

(b) - End-system multicast. Small groups, interactive communication
      (e.g., chat)

    - Overcast. Large groups, streaming large quantities of data
      (e.g., file downloads/updates)

2. Active Networks. (20 pt)

(a) What is the main problem active networks try to address?

(b) Compare in terms of advantages and disadvantages active networks
and overlay networks. Name at least one advantage for each of them.

Answer:

(a) Deploy new protocols and applications at a much faster pace.

(b)
   - Advantages of active networks: efficiency. There is no need to
     tunnel packets, and to process packets at higher layers than the
     network layer.

   - Advantages of overlay networks: easier to deploy.  No need to
     integrate overlay nodes in the network infrastructure. Active
     nodes have to collaborated (be trusted) by the other routers in
     the same AS (they need to exchange routing info)

3. Internet measurements. (20 pt)

(a) Give two reasons of why the default path selection in today's
Internet is not optimal.

Answer:

(a)
1) Intra-domain routing protocols such as RIP and OSPF use # hops
   based metrics which are not necessary correlated to end-to-end
   performances.

2) Inter-domain routing protocols (e.g., BGP) use economic and
   political driven policies which may not take into account the
   end-to-end performances. In fact, transit ISPs have an economic
   incentive to consume minimum resources for the transit traffic,
   which usually results in bad performances for this traffic.