• requirements, terminology
  • nodes, links, flows
  • frames/packets/segments
  • link types: point-to-point, multiple access
  • internetworks
  • (de)multiplexing
  • channel types: streams, datagrams, others?
• building with layers
  • API between layers
  • layering models
    • Internet
    • OSI model (briefly)
  • wireshark interlude
    • packet trace

• end-to-end argument
  • Saltzer, Reed, and Clark, “End-to-End Arguments in System Design” (ACM Transactions on Computing Systems, 1984)
• direct links
  • copper, fiber, radio
• bits as voltage, timing
  • Manchester coding
• types of transmission errors
• framing
  • headers and trailers
  • byte, bits, clocks
• error detection — checksums

• stop-and-wait

• need for sequence numbers

• performance
  • bandwidth, latency, jitter

• sliding window
  • size and bandwidth-delay product
  • sequence number wraparound
  • duplicate ACKs
  • selective ACKs

• TCP realities
  • sequence numbers are bytes
  • TCP options
• switched networks
• packet versus circuit switching
• software defined networks
  • see also: Feamster et al, “The Road To SDN: An Intellectual History of Programmable Networks”

• P4 switch architecture (start)
  • see also: SDN book, chapter 4
  • control versus dataplane
  • match/action pipeline
  • tables
• learning bridges

(Reiss out of town)

The recording is available on Canvas on the Panopto Video tab and will be copied here later.

• IP versus MAC addresses
• broadcast, basic multicast
  • broadcast on ethernet
  • broadcast on IP
• ARP

• ICMPv6 Neighbor Discovery

• IP header
  • checksums

• example: ICMP ping

• autoconfiguration
  • DHCP
  • SLAAC

(Reiss out of town)

The recording is available on Canvas on the Panopto Video tab and will be copied here later.

• graph of throughput versus window size
• history: congestion collapse

• observing good/bad congestion
• steady-state and changing cross-traffic
• slow-increase/fast-decrease
• fair sharing
• additive increase/multiplicative decrease

• min/max fairness
• fast retransmit and fast recovery
• slow start

• quiz review

• queuing theory interlude:
  • randomized arrivals means no 100% utilization
  • high variance from long queues

• bufferbloat: congestion makes queues deep

• timeout management
  • why estimate variance
  • exponential backoff idea
• alternate congestion signals (start)
  • explicit congestion notification
    • echoing back congestion signal

• alternate congestion signals (finish)
  • delay-based congestion control
• queue discplines
  • droptail and problems
  • fair queuing
  • strict priority
  • random early detection

• network as a graph
  • review: routing tables
• ICMP routing errors
  • time-to-live
  • unreachable

• fragments

• spanning tree protocols (start)

• spanning tree protocols (finish)
  • centralized
  • distributed

• routing metrics

• distance vector routing
  • running centrally
  • distributed protocol

• distance vector (con’t)
  • handling link removal
  • count-to-infinity problem
• link state routing
  • running centrally
  • distributed protocol

• autonomous systems
  • graph structure
  • business relationships (customer/provider; content-delivery; peering)

• BGP advertisements

• domain names
  • hierarchy
  • zones

• recursive v authoritative
• record types, format
  • name compression
• DNS questions and answers

• IDNs
• reverse lookups

• preview: cache poisoning, DNSSEC

• BSD sockets
• 5-tuples, netstat
• listening sockets
• conneced/unconnected sockets
• socket / bind / connect
• read/recv, write/send
• server sockets — listen, accept
• recvfrom, sendto, …
• broadcast with sockets
• aside: local sockets

• HTTP/1.1
• URL/URI parts
• request headers, bodies
• request methods (GET, HEAD, POST, …)

• connection keep-alive, pipelining, multiple streams

• HTTP caching
• detecting if up-to-date

• HTTP (finish)
  • HTTP proxies, reverse proxies
  • REST
• carrier sense multiple access
• collision probabilities

• collision detection
• wireles signal propoation
• hidden terminals
• ready-to-send / clear-to-send

• collision detection revisited
• multiple access points
• scheduled wireless networks

• scheduled wireless networks
• firewall placement
• stateless firewalls

• request stuffing
• stateful firewalls
• NAT
• NAT traversal
• IDS

• encapsulation motivations
• “transparent” tunnels/proxy
• encapsulation techniques
• transport-layer proxies
  • SOCKS, Tor (start)

• Tor (finish)
• network-layer tunnels
  • TCP in TCP sidenote
• VLANs
• (if time) MPLS

• notes

2PM-5PM (last exam slot)