1. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science Emery Berger University of Massachusetts Amherst Operating Systems CMPSCI 377 Lecture 19: Network Structures

2. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 2 Next Few Classes Networking basics Distributed services e-mail, www, telnet Distributed operating systems Distributed file systems

3. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 3 Distributed Systems distributed system: set of physically separate processors connected by one or more communication links no shared clock or memory Many systems today distributed in some way e-mail, file servers, network printers, remote backup, web... P2 P1 P3 P4

4. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 4 Parallel vs. Distributed Systems Tightly-coupled systems: “parallel processing” Processors share clock, memory, run one OS Frequent communication Loosely-coupled systems: “distributed computing” Each processor has own memory, runs independent OS Infrequent communication

5. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 5 Advantages of Distributed Systems Resource sharing Computational speedup Reliability Communication

6. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 6 Advantages of Distributed Systems Resource sharing Resources need not be replicated Shared files Expensive (scarce) resources can be shared Color laser printers Processors present same environment to user Keeping files on file server

7. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 7 Advantages, continued Computational speedup n processors = n times computational power SETI@home Problems must be decomposable into subproblems Trivial = embarrassingly parallel Coordination & communication required between cooperating processes Synchronization Exchange of results

8. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 8 Advantages, continued Reliability Replication of resources provides fault tolerance One node crashes, user works on another one Performance degradation but system available Must avoid single point of failure Single, centralized component of system Example: central file servers

9. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 9 Advantages, continued Communication Users/processes on different systems can communicate Mail, transaction processing systems like airlines & banks, www

10. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 10 Distributed Systems Issues Operating systems support for distribution Communication & networks Transparency Security Reliability Performance & scalability Programming models

11. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 11 Networks Goal: provide efficient, correct, robust message passing between two separate nodes Local area network (LAN) – connects nodes in single building, fast & reliable (Ethernet) Media: twisted-pair, coax, fiber Bandwidth: 10-100MB/s Wide area network (WAN) – connects nodes across large geographic area (Internet) Media: fiber, microwave links, satellite channels Bandwidth: 1.544MB/s (T1), 45 MB/s (T3)

12. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 12 Network Topologies Connection of nodes impacts: Maximum & average communication time Fault tolerance Expense Two basic topologies: Point-to-point Bus

13. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 13 Point-to-Point Network Topologies Fully-connected Each message takes one “hop” Node failure – no effect on communication with others Expensive – impractical for WANs

14. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 14 Point-to-Point Network Topologies Partially connected Links between some, but not all nodes Less expensive, less tolerant to failures Single node failure can partition network Sending message takes several hops Needs routing algorithms

15. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 15 Point-to-Point Network Topologies Tree structure: network hierarchy Messages fast between direct descendants Max message cost? Not failure tolerant Any interior node fails – network partitioned

16. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 16 Point-to-Point Network Topologies Star network: all nodes connect to central node Each message takes how many hops? Not failure tolerant Inexpensive – sometimes used for LANs

17. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 17 Point-to-Point Network Topologies One-directional ring Given n nodes, max hops? Inexpensive Fault-tolerant?

18. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 18 Point-to-Point Network Topologies Bi-directional ring Given n nodes, max hops? Inexpensive Fault-tolerant? One node? Two?

19. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 19 Point-to-Point Network Topologies Doubly-connected ring: nodes connected to neighbors & one more distant Given n nodes, max hops? Fault-tolerant? More expensive

20. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 20 Bus Network Topologies Bus nodes connect to common network Linear bus – single shared link Nodes connect directly to each other via bus Inexpensive (linear in # of nodes) Tolerant of node failures Ethernet LAN

21. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 21 Bus Network Topologies Ring bus – single shared circular link Same technology & tradeoffs as linear bus

22. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 22 Principles of Network Communication Data broken into packets Basic unit of transfer Packets sent through network Computers & routers at switching points control packet flow Road analogy: Packets = cars Network = roads Computer = traffic lights (intersection) Too many packets on shared link/node = traffic jam

23. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 23 Communication Protocols Protocol: agreed-upon rules for communication Protocol stack: layers that comprise networking software Each layer N provides service to layer N+1

24. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 24 Traditional Layers Application layer – applications that use the net Presentation layer – data format conversion (big/little endian) Session layer – implements communication strategy (e.g., RPC) Transport layer – reliable end-to-end communication Network layer – routing & congestion control Data link control layer – reliable point-to-point communication over unreliable channel Physical layer – electrical/optical signaling across “wire”

25. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 25 TCP/IP Protocol Stack Internet standard protocol stack TCP: reliable protocol – packets received in order UDP (user datagram protocol) – unreliable No guarantee of delivery

26. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 26 Packet Format Contains all info needed to recreate original message Packets may arrive out of order = need sequence number Data segment contains headers for higher protocol layers & application data

27. U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science 27 Summary Virtually all computer systems contain distributed components Networks connect them Key tradeoffs: Speed Reliability Expense