1. Project3 Malabika Das Prajna Setty Preeti Rao

2. Assumptions and Address Scheme Assumptions End hosts can only connect to one router and knows which router it is connected to. Content copies can be shared between hosts Every router has at most 4 ports New routers can connect to atmost one router in the network All the input ports have input queues and output ports have output queues Total number of contents, hosts and routers- 255 each There is no centralized server Every router requires approx 75kB memory for their routing table Every ARP table requires a memory area of 3.7kB The maximum time taken for a message to reach the farthest host is around 10ms (MaxRTT=20ms)

3. Naming Scheme & Address Scheme End Hosts and Routers- Have a unique MAC address and a unique 9bit network address (511 address spaces + 111111111 for broadcast messages) Host network addresses referred to as H1, H2… H255 Router network addresses referred to as R1, R2…. R255 Every Host is assigned with a network address by the router it is connecting to at startup Content ID- 256 CIDs (serial numbers) available as a list at every host. (C1, C2…. C256) Randomly picks the ID from the table based on an algorithm and assigns to a new content

4. Bootstrapping and Discovery Routers and hosts boot up- tables that need to be initialized

5. Routers discover neighboring routers - Network address is assigned & ARP table is populated and copied to the new router Hosts discovering router: Attached router assigns a network address from its pool of available addresses & informs the host. Periodic “I am alive” message exchange Initialisation of the network

6. ● Failure of a host and revival/ Joining of a new host ○Population of ARP Table of Router ○Request for “CID column” from attached router - BootUp packet ○Updates its content ID table ○Eliminates the need for a new host to learn and update its table over time. ● Failure of a router and revival/ Joining of a new router ○Population of ARP Table of Router ○Request for 2nd and 3rd columns from attached routers - through BootUp packet ○Updates its 2nd column with the used contentIDs. ○Updates third column, by keeping those nodes which are connected directly to it.

7. Baseline Algorithm to choose unique content IDs Content ID table at every host: Contains complete list of available content IDs (001 - 256) Used content IDs are marked as 1; Unused are available for future use; Content files whose copies are present on host are marked with local name of file;

8. Suppose two hosts contend for the same Content_ID.

9. Advertising Content IDs: Generation and Receipt of a content must be advertised/broadcast. announce(contentID, Host address) Routers update their routing tables with the content ID and nearest nodes’(having access to that contentID) network address.

10. Routing of content request packet: At each router: 1.The immediate requesting node is populated against (i.e. prev column) the content ID being requested 2.From the list of immediately available nodes attached to the content, ping to know if the content still available 3.Choose the node with fastest ACK within the timer period. 4.Populate the last column with this node, and stop timer. 5.If ACK received at the same time, choose the one with the lower network address. 6.If No ACK is received within the timer period, perform re-ping. Routing of the content itself: At each router: 1.Already populated table is used 2.Read in the reverse order. 3.Tracing the path back to the requesting host.

11. getContent(C1) { 1.(refer second column of Routing table looking for C1) 2.Get the list of nodes having the content: as in 3rd column of the table; (say H, R2, H3) 3.ping each of those nodes asking if they still have access to the content or not; 4.Keep those nodes which reply with an ACK; (Say R2 and H3) 5.Eliminate the Host from the table after timeout for receipt of ACK; (Say H1) 6.If one node’s ACK reaches before another, consider that as next hop. If 2 ACKs reach at the same time, choose one which has lower network address. No ACK requires re-pinging. 7.Consider H3’s ACk reaches earliest, choose H3 as next hop; 8.H3 loads contents into the DATA packet; 9.Rerouted back using the same routing procedure; 10.Once the requesting host receives the data, it broadcasts to the network the receipt. and accordingly, the data is updated in all the routing tables. }

12. Packet Datagram

15. Packet types for data packet

16. Data Transfer and Reliability ● Message Forward ● Unicast - Our protocol is optimised for the given network for unicast only. ● ARQ Scheme ● In our scheme, we are using Hop-by-hop acknowledgment. ● Stop-and-wait ARQ scheme is used for reliability of the network. Failure to receive the ACK or timeout will cause all the routing tables to be updated

17. Advantages and Disadvantages Scalability - Currently, the network has limited scalability, as the network address space is limited to 512. To add beyond 256 routers to the network, subnetting can be introduced. Latency - Depends on the wait time for the ACK at each router.

18. Example Scenarios Assumption that all links have a weight of 1 H1 H2 H3 C1C2 C3 R1R2R3R4 R5 Scenario 1: @host_H2: get (content_C3)

19. H1 H2 H3 C1 C2 C3 R1R2R3R4 R5 Scenario 2: @host_H1: get (content_C2)

20. H1 C3 Scenario 3: @host_H1: get (content_C1) H2H3 H4 C1 C2 C1