Networks (Distributed computing)

Data Networking & Client-Server Communication Paul Krzyzanowski [email_address] [email_address] Distributed Systems Except as otherwise noted, the content of this presentation is licensed under the Creative Commons Attribution 2.5 License.

Distributed systems Independent machines work cooperatively without shared memory They have to talk somehow Interconnect is the network

Modes of connection Circuit-switched dedicated path guaranteed (fixed) bandwidth [almost] constant latency Packet-switched shared connection data is broken into chunks called packets each packet contains destination address available bandwidth  channel capacity variable latency

What’s in the data? For effective communication same language, same conventions For computers: electrical encoding of data where is the start of the packet? which bits contain the length? is there a checksum? where is it? how is it computed? what is the format of an address? byte ordering

Protocols These instructions and conventions are known as protocols

Protocols Exist at different levels understand format of address and how to compute checksum request web page humans vs. whales different wavelengths French vs. Hungarian versus

Layering To ease software development and maximize flexibility: Network protocols are generally organized in layers Replace one layer without replacing surrounding layers Higher-level software does not have to know how to format an Ethernet packet … or even know that Ethernet is being used

Layering Most popular model of guiding (not specifying) protocol layers is OSI reference model Adopted and created by ISO 7 layers of protocols

OSI Reference Model: Layer 1 Transmits and receives raw data to communication medium. Does not care about contents. voltage levels, speed, connectors Physical 1 Examples: RS-232, 10BaseT

OSI Reference Model: Layer 2 Detects and corrects errors. Organizes data into packets before passing it down. Sequences packets (if necessary). Accepts acknowledgements from receiver. Data Link Physical 1 2 Examples: Ethernet MAC, PPP

OSI Reference Model: Layer 3 Relay and route information to destination. Manage journey of packets and figure out intermediate hops (if needed). Network Data Link Physical 1 2 3 Examples: IP, X.25

OSI Reference Model: Layer 4 Provides a consistent interface for end-to-end (application-to-application) communication. Manages flow control. Network interface is similar to a mailbox. Transport Network Data Link Physical 1 2 3 4 Examples: TCP, UDP

OSI Reference Model: Layer 5 Services to coordinate dialogue and manage data exchange. Software implemented switch. Manage multiple logical connections. Keep track of who is talking: establish & end communications. Session Transport Network Data Link Physical 1 2 3 4 5 Examples: HTTP 1.1, SSL, NetBIOS

OSI Reference Model: Layer 6 Data representation Concerned with the meaning of data bits Convert between machine representations Presentation Session Transport Network Data Link Physical 1 2 3 4 5 6 Examples: XDR, ASN.1, MIME, MIDI

OSI Reference Model: Layer 7 Collection of application-specific protocols Application Presentation Session Transport Network Data Link Physical 1 2 3 4 5 6 7 Examples: email (SMTP, POP, IMAP) file transfer (FTP) directory services (LDAP)

Local Area Network (LAN) Communications network small area (building, set of buildings) same, sometimes shared, transmission medium high data rate (often): 1 Mbps – 1 Gbps Low latency devices are peers any device can initiate a data transfer with any other device Most elements on a LAN are workstations endpoints on a LAN are called nodes

Connecting nodes to LANs ? network computer

Connecting nodes to LANs Adapter expansion slot (PCI, PC Card, USB dongle) usually integrated onto main board Network adapters are referred to as Network Interface Cards ( NIC s) or adapters or Network Interface Component network computer

Media Wires (or RF, IR) connecting together the devices that make up a LAN Twisted pair Most common: STP: shielded twisted pair UTP: unshielded twisted pair (e.g. Telephone cable, Ethernet 10BaseT) Coaxial cable Thin (similar to TV cable) Thick (e.g., 10Base5, ThickNet) Fiber Wireless

Hubs, routers, bridges Hub Device that acts as a central point for LAN cables Take incoming data from one port & send to all other ports Switch Moves data from input to output port. Analyzes packet to determine destination port and makes a virtual connection between the ports. Concentrator or repeater Regenerates data passing through it Bridge Connects two LANs or two segments of a LAN Connection at data link layer (layer 2) Router Determines the next network point to which a packet should be forwarded Connects different types of local and wide area networks at network layer (layer 3)

Networking Topology Bus Network

Networking Topology Tree Network

Networking Topology Star Network

Networking Topology Ring Network

Networking Topology Mesh Network

Transmission networks Baseband All nodes share access to network media on an equal basis Data uses entire bandwidth of media Broadband Data takes segment of media by dividing media into channels (frequency bands)

Broadband: RF broadcasts http://www.ntia.doc.gov/osmhome/allochrt.pdf

Broadband/Baseband: Cable TV 55-552 MHz: analog channels 2-78 553-865 MHz: digital channels 79-136 DOCSIS: Data Over Cable Service Interface Specification (approved by ITU in 1998; DOCSIS 2.0 in 2001) Downstream: 50-750 MHz range, 6 MHz bandwidth - up to 38 Mbps - received by all modems Upstream: 5-42 MHz range - 30.72 Mbps (10 Mbps in DOCSIS 1.0, 1.1) - data delivered in timeslots (TDM) DOCSIS 3.0 features channel bonding for greater bandwidth Broadband Baseband within Broadband +1.25 MHz video +5.75 MHz audio 6 MHz

DOCSIS Modem modulator demodulator tuner MAC CPU network interface Restrictions on upload/download rates set by transferring a configuration file to the modem via TFTP when it connects to the provider. ethernet interface (to PC) cable

Baseband: Ethernet Standardized by IEEE as 802.3 standard Speeds: 100 Mbps - 1 Gbps typical today Ethernet: 10 Mbps Fast Ethernet: 100 Mbps Gigabit Ethernet: 1 Gbps 10 Gbps, 100 Gbps Network access method is Carrier Sense Multiple Access with Collision Detection ( CSMA/CD ) Node first listens to network to see if busy Send Sense if collision occurred Retransmit if collision

Ethernet media Bus topology (original design) originally thick coax (max 500m): 10Base5 then… thin coax (<200m): 10Base2 BNC connector Star topology (central hub or switch) 8 pit RJ-45 connector, UTP cable, 100 meters range 10BaseT for 10 Mbps 100BaseT for 100 Mbps 1000BaseT for 1 Gbps Cables CAT-5: unshielded twisted pair CAT-5e: designed for 1 Gbps CAT-6: 23 gauge conductor + separator for handling crosstalk better

Wireless Ethernet media Wireless (star topology) 802.11 (1-2 Mbps) 802.11b (11 Mbps - 4-5 Mbps realized) 802.11a (54 Mbps - 22-28 Mbps realized) 802.11g (54 Mbps - 32 Mbps realized) 802.11n (108 Mbps - 30-47 Mbps realized) ethernet Access Point

Connecting to the Internet DOCSIS modem via cable TV service DSL router Ethernet converted to ATM data stream Up to 20 Mbps up to ~ 2 km. POTS limited to 300-3400 Hz DSL operates > 3500 Hz Modem Data modulated over voice spectrum (300-3400 Hz) Serial interface to endpoint V.92: 48 kbps downstream, near 56 kbps up Use PPP or SLIP to bridge IP protocol

Connecting to the Internet Dedicated T1 or T3 line T1 line: 1.544 Mbps (24 PCM TDMA speech lines @ 64 kbps) T3 line: 44.736 Mbps (672 channels) CSU/DSU at router presents serial interface Channel Service Unit / Data Service Unit CSU/DSU T1 line RS-232C, RS-449, V.xx serial line router LAN Phone network

Connecting to the Internet Fiber to the Home, Fiber to the Curb Ethernet interface E.g., Verizon’s FiOS 30 Mbps to the home Long Reach Ethernet (LRE) Ethernet performance up to 5,000 feet Wireless: WiMax EDGE (70-135 Kbps) GPRS (<32 Kbps)

Clients and Servers Send messages to applications not just machines Client must get data to the desired process server process must get data back to client process To offer a service, a server must get a transport address for a particular service well-defined location

Machine address versus Transport address

Transport provider Layer of software that accepts a network message and sends it to a remote machine Two categories: connection-oriented protocols connectionless protocols

Connection-oriented Protocols establish connection [negotiate protocol] exchange data terminate connection

Connection-oriented Protocols virtual circuit service provides illusion of having a dedicated circuit messages guaranteed to arrive in-order application does not have to address each message vs. circuit-switched service establish connection [negotiate protocol] exchange data terminate connection dial phone number [decide on a language] speak hang up analogous to phone call

Connectionless Protocols - no call setup - send/receive data (each packet addressed) - no termination

Connectionless Protocols datagram service client is not positive whether message arrived at destination no state has to be maintained at client or server cheaper but less reliable than virtual circuit service - no call setup - send/receive data (each packet addressed) - no termination drop letter in mailbox (each letter addressed) analogous to mailbox

Ethernet Layers 1 & 2 of OSI model Physical (1) Cables: 10Base-T, 100Base-T, 1000Base-T, etc. Data Link (2) Ethernet bridging Data frame parsing Data frame transmission Error detection Unreliable, connectionless communication

Ethernet 48-byte ethernet address Variable-length packet 1518-byte MTU 18-byte header, 1500 bytes data Jumbo packets for Gigabit ethernet 9000-byte MTU dest addr src addr frame type 6 bytes 6 bytes 2 data (payload) CRC 4 46-1500 bytes 18 bytes + data

IP – Internet Protocol Born in 1969 as a research network of 4 machines Funded by DoD’s ARPA Goal : build an efficient fault-tolerant network that could connect heterogeneous machines and link separately connected networks.

Internet Protocol Connectionless protocol designed to handle the interconnection of a large number of local and wide-area networks that comprise the internet IP can route from one physical network to another

IP Addressing Each machine on an IP network is assigned a unique 32-bit number for each network interface: IP address , not machine address A machine connected to several physical networks will have several IP addresses One for each network

IP Address space 32-bit addresses  >4 billion addresses! Routers would need a table of 4 billion entries Design routing tables so one entry can match multiple addresses hierarchy: addresses physically close will share a common prefix

IP Addressing: networks & hosts first 16 bits identify Rutgers external routers need only one entry route 128.6.*.* to Rutgers cs.rutgers.edu 128.6.4.2 80 06 04 02 remus.rutgers.edu 128.6.13.3 80 06 0D 03 network # host #

IP Addressing: networks & hosts IP address network #: identifies network machine belongs to host #: identifies host on the network use network number to route packet to correct network use host number to identify specific machine

IP Addressing Expectation: a few big networks and many small ones create different classes of networks use leading bits to identify network To allow additional networks within an organization: use high bits of host number for a “network within a network” – subnet class leading bits bits for net # bits for host A 0 7 (128) 24 (16M) B 10 14 (16K) 16 (64K) C 110 21 (2M) 8 (256)

IP Addressing IBM: 9.0.0.0 – 9.255.255.255 0 0001001 xxxxxxxx xxxxxxxx xxxxxxxxx network # 8 bits host # 24 bits 0 0001001 10101010 11 xxxxxx xxxxxxxxx network # 18 bits host # 14 bits Subnet within IBM (internal routers only)

Running out of addresses Huge growth Wasteful allocation of networks Lots of unused addresses Every machine connected to the internet needed a worldwide-unique IP address Solutions: CIDR, NAT, IPv6

Classless Inter-Domain Routing (CIDR) Replace class A, B, C addresses: Explicitly specify # of bits for network number rather than 8 (A), 16 (B), 24 (C) bits Better match for organizational needs machine that needs 500 addresses: get a 23-bit network number (512 hosts) instead of a class B address (64K hosts)

Classless Inter-Domain Routing How does a router determine # bits? CIDR address specifies it: 32-bit-address / bits-for-network-prefix 128.6.13.3/16 /27 : 1/8 of a class C (32 hosts) /24 : class C /16 : class B managing CIDR addresses & prefixes can be a pain

IP Special Addresses All bits 0 Valid only as source address “ all addresses for this machine” Not valid over network All host bits 1 Valid only as destination Broadcast to network All bits 1 Broadcast to all directly connected networks Leading bits 1110 Class D network 127.0.0.0 : reserved for local traffic 127.0.0.1 usually assigned to loopback device

IPv6 vs. IPv4 IPv4 4 byte (32 bit) addresses IPv6: 16-byte (128 bit) addresses 3.6 x 10 38 possible addresses 8 x 10 28 times more addresses than IPv4 4-bit priority field Flow label (24-bits)

Network Address Translation (NAT) External IP address 24.225.217.243 Internal IP address 192.168.1.x .1 .2 .3 .4 .5

Getting to the machine IP is a logical network on top of multiple physical networks OS support for IP: IP driver IP driver network driver send data send packet to wire from wire receive packet receive data

IP driver responsibilities Get operating parameters from device driver Maximum packet size (MTU) Functions to initialize HW headers Length of HW header Routing packets From one physical network to another Fragmenting packets Send operations from higher-layers Receiving data from device driver Dropping bad/expired data

Device driver responsibilities Controls network interface card Comparable to character driver Processes interrupts from network interface Receive packets Send them to IP driver Get packets from IP driver Send them to hardware Ensure packet goes out without collision bottom half top half

Network device Network card examines packets on wire Compares destination addresses Before packet is sent, it must be enveloped for the physical network device header payload IP header IP data

Device addressing IP address  ethernet address Address Resolution Protocol (ARP) Check local ARP cache Send broadcast message requesting ethernet address of machine with certain IP address Wait for response (with timeout)

Routing Router Switching element that connects two or more transmission lines (e.g. Ethernet) Routes packets from one network to another (OSI layer 2) Special-purpose hardware or a general-purpose computer with two or more network interfaces

Routing Packets take a series of hops to get to their destination Figure out the path Generate/receive packet at machine check destination If destination = local address, deliver locally else Increment hop count (discard if hop # = TTL) Use destination address to search routing table Each entry has address and netmask. Match returns interface Transmit to destination interface Static routing

Dynamic Routing Class of protocols by which machines can adjust routing tables to benefit from load changes and failures Route cost: Hop count (# routers in the path) Time: Tic count – time in 1/18 second intervals

Dynamic Routing Examples RIP (Routing Information Protocol) Exchange routing tables with neighboring routers on internal networks Choose best route if multiple routes exist OSPF (Open Shortest Path First) Tests status of link to each neighbor. Sends status info on link availability to neighbors. Cost can be assigned on reliability & time BGP (Border Gateway Protocol) TCP connection between pair of machines Route selection based on distance vector Exchanges information about reachable networks Periodic keep-alive messages

Transport-layer protocols over IP IP sends packets to machine No mechanism for identifying sending or receiving application Transport layer uses a port number to identify the application TCP – Transmission Control Protocol UDP – User Datagram Protocol

TCP – Transmission Control Protocol Virtual circuit service (connection-oriented) Send acknowledgement for each received packet Checksum to validate data Data may be transmitted simultaneously in both directions

UDP – User Datagram Protocol Datagram service (connectionless) Data may be lost Data may arrive out of sequence Checksum for data but no retransmission Bad packets dropped

IP header device header payload IP header IP data TCP/UDP header total length options and pad svc type (TOS) vers hlen fragment identification flags fragment offset TTL protocol header checksum source IP address destination IP address 20 bytes

Headers: TCP & UDP device header IP header IP data TCP/UDP header TCP header UDP header 20 bytes 8 bytes payload src port dest port seq number ack number hdr len flags checksum urgent ptr options and pad - window src port dest port seg length checksum

Device header (Ethernet II) device header IP header IP data TCP/UDP header dest addr src addr frame type 6 bytes 6 bytes 2 data CRC 4 46-1500 bytes 18 bytes + data payload

Quality of Service Problems in IP Too much traffic Congestion Inefficient packet transmission 59 bytes to send 1 byte in TCP/IP! 20 bytes TCP + 20 bytes IP + 18 bytes ethernet Unreliable delivery Software to the rescue – TCP/IP Unpredictable packet delivery

IP Flow Detection Flow detection in routers: Flow : set of packets from one address:port to another address:port with same protocol Network controls flow rate by dropping or delaying packets With flow detection: drop TCP packets over UDP Discard UDP flow to ensure QoS for other flows With flow detection: Traffic Shaping Identify traffic flows Queue packets during surges and release later High-bandwidth link to low-bandwidth link Traffic Policing Discard traffic that exceeds allotted bandwidth

Dealing with congestion FIFO queuing Priority queues Flow-based weighted fair queuing Group all packets from a flow together Class-based weighted fair queuing Based on protocols, access control lists, interfaces, etc. Custom queues

Inefficient Packets Lots of tiny packets Head-of-line blocking Nagle’s algorithm: buffer new data if unacknowledged data outstanding Header/packet compression Link-to-link Header compression (RFC 3843) Payload compression (RFC 2393) $ delivery vs. $ compression

Differentiated Services (soft QoS) Some traffic is treated better than others Statistical - no guarantees TOS bits & Diff-Serv Use on Internet is limited due to peering agreement complexities

TOS bits Advisory tag in IP header for use by routers TOS: Type Of Service , 4 bits Minimum Delay [0x10] FTP , telnet, ssh Maximum Throughput [0x08] ftp-data, www Maximum reliability [0x04] SNMP , DNS Minimum cost [0x02] NNTP , SMTP RFC 1349, July, 1992

Differentiated Services (Diff-Serv) Revision of interpretation of ToS bits ToS field in IP header Differentiated Sevices Control Point (DSCP) p p p d t r - - Priority: 0-7 Reliability: normal/high Throughput: normal/high Delay: normal/low RFC 2475, December 1998

Guaranteed QoS (hard QoS) Guarantee via end-to-end reservation

Reservation & Delivery Protocol RSVP: ReSerVation Protocol Hosts request specific quality of service Routers reserve resources RFC 2205

Media Delivery Protocols Real-Time Control Protocol (RTCP) Provides feedback on QoS (jitter, loss, delay) RFC 3550 RTP: Real-Time Transport Protocol Not a routing protocol No service guarantees Provides: Payload identification sequence # time stamp RTP/RTCP do not provide QoS controls

ATM: Asynchronous Transfer Mode Late 1980’s Goal: Merge voice & data networking low but constant bandwidth high but bursty bandwidth

ATM Traditional voice networking Circuit switching Too costly Poor use of resource Does not lend to multicasting ATM Based on fixed-size packets over virtual circuits Fixed-size cells provide for predictive scheduling Large cells will not hold up smaller ones Rapid switching

ATM Current standard: 53-byte cell: 48-byte data, 5-byte header Sender specifies traffic type upon connecting: CBR Constant bit-rate bandwidth Uncompressed video, voice VBR Variable bit-rate Avg, peak bandwidth Compressed video, voice ABR Available bit-rate -none- ftp, web access

ATM Small cells  lots of interrupts >100,000/second ATM hardware supports an ATM Adaptation Layer (AAL) Converts cells to variable-sized (larger) packets: AAL 1: for CBR AAL 2: for VBR AAL 3/4: ABR data AAL 5: ABR data, simplified AAL 6: MPEG-2 video

Sockets IP lets us send data between machines TCP & UDP are transport layer protocols Contain port number to identify transport endpoint (application) One popular abstraction for transport layer connectivity: sockets Developed at Berkeley

Sockets Attempt at generalized IPC model Goals: communication between processes should not depend on whether they are on the same machine efficiency compatibility support different protocols and naming conventions

Socket Abstract object from which messages are sent and received Looks like a file descriptor Application can select particular style of communication Virtual circuit, datagram, message-based, in-order delivery Unrelated processes should be able to locate communication endpoints Sockets should be named Name meaningful in the communications domain

Step 1 Create a socket int s = socket(domain, type, protocol) AF_INET SOCK_STREAM SOCK_DGRAM useful if some families have more than one protocol to support a given service

Step 2 Name the socket (assign address, port) int error = bind(s, addr, addrlen) socket Address structure struct sockaddr* length of address structure

Step 3a (server) Set socket to be able to accept connections int error = listen(s, backlog) socket queue length for pending connections

Step 3b (server) Wait for a connection from client int snew = accept(s, clntaddr, &clntalen) socket pointer to address structure length of address structure new socket for this session

Step 3 (client) Connect to server int error = connect(s, svraddr, svraddrlen) socket Address structure struct sockaddr* length of address structure

Step 4 Exchange data Connection-oriented read/write recv/send (extra flags) Connectionless sendto , sendmsg recvfrom , recvmsg

Step 5 Close connection shutdown(s, how ) how : 0: can send but not receive 1: cannot send more data 2: cannot send or receive (=0+1)

Sockets in Java java.net package Two major classes: Socket : client-side ServerSocket : server-side

Step 1a (server) Create socket and name it ServerSocket svc = new ServerSocket( port )

Step 1b (server) Wait for connection from client Server req = svc.accept() new socket for client session

Step 1 (client) Create socket and name it Socket s = new Socket( address , port); obtained from: getLocalHost, getByName, or getAllByName Socket s = new Socket( “cs.rutgers.edu” , 2211);

Step 2 Exchange data obtain InputStream/OutputStream from Socket object BufferedReader in = new BufferedReader( new InputStreamReader( s.getInputStream())); PrintStream out = new PrintStream(s.getOutputStream());

Step 3 Terminate connection close streams, close socket in.close(); out.close(); s.close();

Protocol Control Block Client only sends data to {machine, port} How does the server keep track of simultaneous sessions to the same {machine, port}? OS maintains a structure called the Protocol Control Block (PCB)

Server: svr=socket() Create entry in PCB table Client Server svr Local addr Local port Foreign addr Foreign port L? Local addr Local port Foreign addr Foreign port L?

Server: bind(svr) Assign local port and address to socket bind(addr=0.0.0.0, port=1234) Client Server svr Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 Local addr Local port Foreign addr Foreign port L?

Server: listen(svr, 10) Set socket for listening Client Server svr Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * Local addr Local port Foreign addr Foreign port L?

Server: snew=accept(svr) Block – wait for connection Client Server svr Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * Local addr Local port Foreign addr Foreign port L?

Client: s=socket() Create PCB entry Client Server svr s Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * Local addr Local port Foreign addr Foreign port L?

Client: s=bind(s) Assign local port and address to socket bind(addr=0.0.0.0, port=7801) Client Server svr s Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * Local addr Local port Foreign addr Foreign port L? 0.0.0.0 7801

Client: connect(s) Send connect request to server [135.250.68.3:7801] to [192.11.35.15:1234] Client Server svr s snew Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * 192.11.35.15 1234 135.250.68.3 7801 Local addr Local port Foreign addr Foreign port L? 0.0.0.0 7801

Client: connect(s) Server responds with acknowledgement [192.11.35.15:1234] to [135.250.68.3 :7801] Client Server svr s snew Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * 192.11.35.15 1234 135.250.68.3 7801 Local addr Local port Foreign addr Foreign port L? 0.0.0.0 7801 192.11.35.15 1234

Communication Each message from client is tagged as either data or control (e.g. connect ) If data – search through table where FA and FP match incoming message and listen =false If control – search through table where listen =true Server svr snew Local addr Local port Foreign addr Foreign port L? 0.0.0.0 1234 * 192.11.35.15 1234 135.250.68.3 7801

Networks (Distributed computing)

Recommended

More Related Content

What's hot (20)

Similar to Networks (Distributed computing) (20)

More from Sri Prasanna (20)

Recently uploaded (20)

Networks (Distributed computing)