1. In a Shared Memory Architecture, multiple processors can read and write data from the same location.
Message Passing Architecture allows an alternate method for communication and movement of data among multiprocessors (compared to shared memory multiprocessor systems): There is no global memory so it provides a way to move data from one local memory to another by means of message passing.
Give a comparison between shared memory and message passing: where are they different and where are the similar? (500 words)
2. Most of the approaches to building a high-speed router architecture recommend implementing the Address Resolution Protocol (ARP) functionality in the slow path. Cite and explain some reasons why this functionality should be in the fast path and not in the slow path.
(300 words)
3. When a layer 2 switch receives a broadcast packet, it floods the broadcast out on all its associated ports. This broadcast traffic can create congestion and bring the overall efficiency and performance of the network down. In such a situation, can the use of a Layer 3 switch help mitigate this problem to some extent?
4. A set of rules is shown in the table below. You are required to construct a Lucent Bit Vector data structure for this set of rules. You must use a bitmap of size 8 bits for characterizing the rules. In your answer state how many memory accesses will be required for classifying a packet with F1 = 0011 and F2 = 1111. You need not show the tries corresponding to the rule matches.
5. Consider an 8×8 hypercube network. Using the E-cube routing algorithm, show the sequence of steps to route messages from node (101101) to node (011010). You must identify all the intermediate nodes on this routing path.
5 questions,
question 1= 500
question 2= 300
question 3 = 250
question 4&5= mathematical algorithm no word limit
This assignment has been designed to test in the following areas:
#Layer 2/3 switches
#Router architecture
# Interconnections
# Classification algorithms