2 (a)
LAN: Local Area Network
1) Low Bandwidth
2) Low Latency (RTT)
WAN: Wide Area Network (leased lines, etc)
1) High Bandwidth
2) High Latency (RTT)
TCP Window Size can be good for both - adapts to Bandwidth + Latency
Timeouts are also adaptive!
LAN over TCP
=======
Good:
Bad: Under congestion, TCP can break down
WAN over TCP
=========
Good:
Bad: TCP Window Size can be too small, Expensive Handshake for TCP?
2 (b)
Fast Retransmit:
a) Instead of retransmitting on timeouts, we *also* retransmit on recv 3 duplicate ACKs
b) Good: Recover packet losses without waiting for the timeout
c) Does Not Help: If there are a lot of re-orderings, we retransmit packets we don't have to
Selective ACK:
a) ACK the exact start/end of the bytes received (so you can ACK exactly what you have)
b) Good: Sender can only retransmit what is missing (esp when sending packets are expensive)
c) Does Not Help: If packet losses are low, SACKs form alot of overhead. (any more suggestions?)
===============================
3) (a)
Congestion Collapse:
Congestive collapse (or congestion collapse) is a condition which a packet switched computer network can reach, when little or no useful communication is happening due to congestion. When a network is in such a condition, it has settled (under overload) into a stable state where traffic demand is high but little useful throughput is available, and there are high levels of packet delay and loss (caused by routers discarding packets because their output queues are too full).
3 (b)
True. Each endsystem implements its own flavor of TCP/IP.
Alternative: Use routers with QoS. Control the internet service providers, end net neutrality.
QoS = Quality of Service => Packets are discriminated based on their content (e.g. discard 50% of bittorrent if traffic gets too high and keep all SSH packets).
Comments (0)
You don't have permission to comment on this page.