202 Lecture Summaries
Slides in this set (19)
-
- - Friendly Names
- - IP Addresses
- - Application Ports
-
- - The InetAddress class deals with resolving names to addresses
- - The Socket class represents an individual connection established over the network
- - The ServerSocket class is a special type of socket used to allow new connections to be made
- - The PrintWriter class allows us to write individual lines of text over a socket
- - The BufferedReader class allows us to read lines of text from a socket
-
- - Methods that block are a problem for communications systems as they prevent other clients getting service
-
- • OSI 7 layer model
- • Encapsulation of information to form frames
- • Defined some physical characteristics of Ethernet
- • Wireless Ethernets, and issues of them
-
- Ethernet
- • Frames
- • MAC addresses
- • Network Interface Cards
- • Hubs, switches
- • IP • Packets
- • IPv4 addresses
- • Routers
-
- • ARP maps IPv4 addresses to Ethernet MAC addresses
- • NDP maps IPv6 addresses to Ethernet MAC addresses
- • ARP/NDP tables are dynamically managed
- • ARP/NDP tables are scoped to a single Ethernet bus
- • Standards documents (RFCs):
- • RFC 826: IPv4 ARP
- • RFC 4861: IPv6 Neighbor Discovery
-
- • IP addresses contain a built-in network id
- • Useful scaling property: can often aggregate smaller network IDs into a single larger network ID.
- • Extract network ID by masking it out
- • Systems determine how to reach a destination by consulting their IP routing table
-
- • IP Routing tables are organised so that the most specific route is chosen to a destination
- • Internet routing is organised using the BGP protocol
- • ASes
- • An AS being a network operated by a single organisation
- • In general, an AS prefers the highest preference route available, then the one with the shortest AS path length. Other rules follow, but these are two fundamental rules
-
- • DNS is an application-level service that provides the ability to obtain an IP address from a more easily remembered name
- • DNS is implemented as a distributed hierarchical database
- • DNS has built-in redundancy, which the DNS application service takes advantage of
- • UDP is an unreliable transport protocol
-
- • DNS is an application-level service that provides the ability to obtain an IP address from a more easily remembered name
- • DNS is implemented as a distributed hierarchical database
- • DNS has built-in redundancy, which the DNS application service takes advantage of
- • UDP is an unreliable transport protocol
-
- •TCP is a transport protocol that is:
- • Reliable
- • Unacknowledged packets are retransmitted after they are lost and/or time out
- • connection-oriented,
- • Hosts establish connection with 3-packet handshake
- • tear-down connections with 4-packet exchange
- • stream-based • Client socket sees a stream of bytes, like reading or writing a file.
- • Does not see individual packets, or retransmitted packets, or connection establishment/teardown packets.
-
- • TCP is the most widely used transport protocol in the Internet: reliable, byte-stream, connection oriented,
- • and FAIR!
- • TCP has to work on paths that vary with length (delay) and available capacity
- • TCP retransmit timer: decide when to declare a packet lost
- • Slow-start: quickly infer the rate at which TCP should be transmitting at
- • Congestion avoidance: slowly increase the rate at which new data is sent to avoid causing congestion
- • Fast retransmit / fast recovery: recover quickly from single packet loss
-
- • Dynamic routing protocols are useful
- • Less tedious than manually reconfiguring all routers when links change (are added, removed, or otherwise change distance)
- • Less error prone
- • Better at responding to failures
- • OSPF is one example of an internal routing protocol
- • Link-state
- • Uses Dijkstra’s algorithm
- • Multicast: send data once to all interested receivers
- • OSPF uses multicast to send routing information to all routers it is connected to.
-
- Encryption allows us to protect sensitive data traversing untrusted networks
- • One-way hashes and public key encryption allow us to build digital signatures
- • Authentication systems allow us to provide our identity across untrusted networks
- • Firewalls allow us to police traffic entering and exiting a network
-
- • The TLS protocol provides a secure way to establish a symmetric key to encrypt communications
- • Protocol relies on certificate verification:
- • Is there a valid certificate path to a CA?
- • Does the hostname match that found in the certificate presented?
-
- • Goal of today’s lecture was to discuss how to use the Java JSSE APIs to do TLS
- • Other APIs in other languages are similar. Use TLS libraries with care. Know what your responsibilities are.
- • JSSE will become more concrete in assignment five
-
- • It became obvious fairly on in the early 1990s that the Internet was going to run out of addresses
- • CIDR, Private addresses bought us time
- • NAT is the price we paid
- • IPv6 is intended solution
- • Increasing deployment, has been defined for over 15 years (1998)
- • Service provider NAT not a good solution.
- • Deployment issues mostly at the edge of the network: enterprise and access networks.
- • Most large transit providers have deployed IPv6
-
- • Today’s lecture was on elementary network debugging
- • Ping basic statistics on a path to a destination
- • Traceroute arrangement of routers on forward path
- • Both use ICMP
- • Try using ping and traceroute in the lab
-
- UDP VS TCP
Information
- Description
- Slide Set on 202 Lecture Summaries, created by Cherese Eriepa on 20/10/2017.
- Created by
- Cherese Eriepa
- Views
- 3
- Rating