This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: ECSE 414 – Intro to Telecom Networks – Fall 2009 Midterm Exam Question 1 [30 marks] a) • The telephone network uses a circuit switched architecture, and the internet uses a packet switched architecture. • In the telephone network, the intelligence resides in the network core (switches). In the internet, the intelligence is primarily in the end-hosts. b) (Answers should include three of the following) • An advantage of statistical multiplexing is that resources don’t get wasted if a connection is idle. • An advantage of statistical multiplexing is that there is no network-level connection setup, which causes additional overhead. • An advantage of statistical multiplexing is that routers do not need to store state in- formation about the connections that are passing through them. So if the router goes down (e.g., looses power), the connections that were using the router are not necessarily dropped. • A disadvantage of statistical multiplexing is that each connection is not guaranteed a minimum bandwidth. • A disadvantage of statistical multiplexing is that the network can become congested if too many users are trying to send packets at the same time, and congestion leads to larger delays and loss. c) 1) Content that is stored in the local cache is available immediately, so the delay for the content is reduced dramatically. 2) Also, because fewer requests must go out the internet, there is less congestion on the access link and lower access delay. d) Root DNS servers receive requests relatively infrequently because DNS requests are cached at top-level domain and local DNS servers. e) A TCP flow or connection is defined by the sender IP address, sender port, destination IP address, and destination port. f) The bandwidth-delay product is the product of the bandwidth available to a connection and the round-trip-time delay for that connection. It has units of bits or bytes. g) If I operated a network at my cottage up in the woods and my local DNS server had a slow connection to the rest of the internet, I would prefer to use recursive DNS queries because the delay from my network to other servers would probably be larger (slow connection) than the delay from other DNS servers involved in the query (root or top-level domain) to the authoritative server. 1 h) In circular distributed hash tables, the peer responsible for a given key is the immediate successor or peer with the next largest value. If there are peers at locations 2, 3, 7, 11, 18, 22, and 25, then key 8 will be stored at peer 11. Because the DHT is circular, the key values wrap around, and key 29 will be with the peer at location 2....
View Full Document
- Fall '10
- Transmission Control Protocol, transmission delay, Single PersisTenT connecTion, nack nseg