06-Dec-17. Credits:4. Notes by Pritee Parwekar,ANITS 06-Dec-17 1

Transcription

1 Credits:4 1 Understand the Distributed Systems and the challenges involved in Design of the Distributed Systems. Understand how communication is created and synchronized in Distributed systems Design and Implement Distributed applications using Technologies like RPC, threads. Learn how to store data in Distributed File System. Understand How Distributed Shared Memory is managed. Notes by Pritee Parwekar,ANITS 06-Dec

2 Introduction to Distributed Systems, What is a Distributed System?, Hard ware concepts, Software concepts, Design issues. Communication in Distributed Systems, Lay red Protocols, ATM networks, The Client sever model, Remote Procedure call, Group communication. Synchronization in Distributed System, Clock Synchronization, Mutual Exclusion, Election algorithms, Atomic transactions, Deadlocks in Distributed Systems. Process and processors in Distributed System threads, System Models, Processors allocation, Scheduling in Distributed System, Fault tolerance, Real time Distributed System. Distributed File Systems, Distributed File System Design, Distributed File System implementation, Trends in Distributed File System. Distributed Shared Memory, Introduction, What is Shared memory?, Consistency models, Page based Distributed Shared memory, Shared variable Distributed Shared memory, Object based Distributed Shared Memory. 3 TEXT BOOK: Distributed Operating Systems, Andrew S. Tanenbaum REFERENCE BOOK: Advanced Concepts in Operating Systems, Makes Singhal and Niranjan G.Shivaratna. 4 2

3 To understand the function of Operating Systems (OS) and what are the types of operating systems. To make the students learn about the concept of Distributed Operating Systems (DOS). 5 By end of this lecture, the students would know the following:- What are operating systems? What are the functions of operating systems? Types and operating systems Functions of operating systems Introduction to distributed operating systems Difference between distributed and network operating systems Examples of operating systems 6 3

4 7 8 4

5 Batch Operating Systems Timesharing Operating Systems Real Time Systems Network Operating Systems (NOS) Distributed Operating Systems (DOS)

6 The users of a batch operating system do not interact with the computer directly. Each user prepares his job on an off-line device like punch cards and submits it to the computer operator. To speed up processing, jobs with similar needs are batched together and run as a group. The programmers leave their programs with the operator and the operator then sorts the programs with similar requirements into batches

7 Time-sharing is a technique which enables many people, located at various terminals, to use a particular computer system at the same time. Time-sharing or multitasking is a logical extension of multiprogramming. Processor's time which is shared among multiple users simultaneously is termed as time-sharing. 13 A real-time system is defined as a data processing system in which the time interval required to process and respond to inputs is so small that it controls the environment. The time taken by the system to respond to an input and display of required updated information is termed as the response time. So in this method, the response time is very less 14 7

8 Hard real-time systems guarantee that critical tasks complete on time. In hard real-time systems, secondary storage is limited or missing and the data is stored in ROM. In these systems, virtual memory is almost never found. 15 Extremely Fast Reaction Time 16 8

9 Soft real-time systems are less restrictive. A critical real-time task gets priority over other tasks and retains the priority until it completes. Soft real-time systems have limited utility than hard real-time systems. For example, multimedia, virtual reality, Advanced Scientific Projects like undersea exploration and planetary rovers, etc. 17 Multimedia / Sending & Receiving Packets 18 9

10 A Network Operating System runs on a server and provides the server the capability to manage data, users, groups, security, applications, and other networking functions. The primary purpose of the network operating system is to allow shared file and printer access among multiple computers in a network, typically a local area network (LAN), a private network or to other networks

11 Centralized servers are highly stable. Security is server managed. Upgrades to new technologies and hardware can be easily integrated into the system. Remote access to servers is possible from different locations and types of systems. 21 High cost of buying and running a server. Dependency on a central location for most operations. Regular maintenance and updates are required

12 Distributed Operating System is a model where distributed applications are running on multiple computers linked by communications. Distributed system is a collection of independent computers that appear to the user of the system as a single computer. A distributed operating system is an extension of the network operating system that supports higher levels of communication and integration of the machines on the network

13 Un-reliability of communication Lack of global knowledge Lack of synchronization & casual ordering Concurrency control Failure of recovery 25 Network Operating Systems :- Contains N copies of Operating Systems, communication between machines is via shared files. Distributed OS : - Contains N copies of Operating systems, communication between nodes is via messages over a network. These messages pass the necessary parameters for the task and on completion messages return the results. Its as if computers sends s to other computers with request and answer

14 27 Resource Sharing High Performance Scalability Fault Tolerance 28 14

15 Give more performance than single system If one pc in distributed system malfunction or corrupts then other node or pc will take care of More resources can be added easily. Resources like printers can be shared on multiple pc s 29 Item Economics Speed Inherent Distribution Reliability Incremental Growth Description Microprocessors offer a better price/performance than mainframes A distributed system may have more total computing power than a mainframe Some applications involve spatially separated machines If one machine crashes, the system as a whole can still survive Computing power can be added in small increment 30 15

16 Item Data Sharing Device Sharing Communication Flexibility Description Allow many users access to a common data base Allow many users to share expensive peripherals like color printers Make human-to-human communication easier, for example, by electronic mail Spread the workload over the available machines in the most effective way 31 Security problem due to sharing, easy access also applies to secret data Some messages can be lost in the network system. Bandwidth is another problem if there is large data then all network wires to be replaced which tends to become expensive Software : Little software exists at present for distributed systems Networking : Network can saturate or cause other problems 32 16

17 In this lecture we have learnt What are operating systems? What are the functions of operating systems? Types and operating systems Functions of all types of operating systems Introduction to distributed operating systems Difference between distributed and network operating systems Examples of operating systems

18 To understand the hardware and software concepts of distributed operating concepts To understand the design issues comes with hardware and software concepts 35 By end of this lecture, the students would know the following:- What is the term multiprocessor? What is the term multicomputer? Flynn s Classification How multiple processors are connected? How multiple processors communicate? What are the hardware and software concepts? Design issues involved 36 18

19 All Distributed Systems consist of multiple CPUs and there are different ways of interconnecting them and how they communicate 37 SISD MISD SIMD MIMD 38 19

20 MIMD (Multiple-Instruction Multiple-Data) MIMD can be split into two classifications Multiprocessors - CPUs share a common memory Multicomputer - CPUs have separate memories 39 Tightly-coupled - short delay in communication between computers, high data rate (e.g., Parallel computers working on related computations) Loosely-coupled - Large delay in communications, Low data rate (Distributed Systems working on unrelated computations) 40 20

21 Bus - All machines connected by single medium (e.g., LAN, bus, backplane, cable) Switched - Single wire from machine to machine, with possibly different wiring patterns (e.g, Internet)

22 Up to 64 Tightly-coupled CPUs on a bus, backplane or motherboard with a shared memory module Plus point is - coherent Disadvantage is Bus traffic So add high speed memory cache to each CPU Snoopy cache

23 45 To build a multiprocessor with more than 64 processors, a different method is needed to connect the CPUs with the memory. Two switching techniques are employed for it. a. Crossbar Switch b. Omega Switch 46 23

24 47 Memory is divided into the modules and are connected to the CPUs with the crossbar switch. At every intersection is a tiny electronic crosspoint switch that can be opened and closed in hardware. When a CPU wants to access a particular memory, the crosspoint switch connecting them is closed, to allow the access to take place. If two CPUs try to access the same memory simultaneously, one of them will have to wait. The downside of the crossbar switch is that with n CPUs and n memories, n2 crosspoint switches are needed. For large n this number can be prohibitive

25 49 NUMA (Non-Uniform Memory Access): placement of program and data building a large, tightly-coupled, shared memory multiprocessor is possible, but is difficult and expensive 50 25

26 It contains 2x2 switches, each having two inputs and two outputs. Each switch can route either input to either output. A careful look at the figure will show that with proper setting of the switches, every CPU can access every memory. In general case, with n CPUs and n memories, the omega network requires log2n switching stages, each containing n/2 switches, for a total of (nlog2n)/2 switches. 51 Bus-Based Multicomputers easy to build communication volume much smaller relatively slow speed LAN ( MIPS, compared to 300 MIPS and up for a backplane bus) Switched Multicomputers interconnection networks: E.g., grid, Hypercube Hypercube: n-dimensional cube bus) 52 26

27 53 Software more important for users Three types: 1. Network Operating Systems 2. (True) Distributed Systems 3. Multiprocessor Time Sharing 54 27

28 loosely-coupled software on loosely-coupled hardware A network of workstations connected by LAN each machine has a high degree of autonomy rlogin machine rcp machine1:file1 machine2:file2 Files servers: client and server model Clients mount directories on file servers Best known network OS: Sun s NFS (network file servers) for shared file systems a few system-wide requirements: format and meaning of all the messages exchanged

29 NFS Architecture Server exports directories Clients mount exported directories NSF Protocols For handling mounting For read/write: no open/close, stateless 57 tightly-coupled software on loosely-coupled hardware provide a single-system image or a virtual uniprocessor a single, global interprocess communication mechanism, process management, file system; the same system call interface everywhere Ideal definition: A distributed system runs on a collection of computers that do not have shared memory, yet looks like a single computer to its users

30 Tightly-coupled software on tightly-coupled hardware Examples: high-performance servers shared memory single run queue traditional file system as on a single-processor system: central block cache

31 Item N/W OS DISTRIBUTED OS MULTIPROCESSOR TIME SHARING OS Does it look like virtual uniprocessor? Do all have to run the same OS? How many copies of OS are there? How is communication achieved? Are you agreed on N/W protocol required? Is there a single run queue? Dose file have well defined semantics?

32 Transparency Flexibility Reliability Performance Scalability 63 How to achieve the single-system image, i.e., how to make a collection of computers appear as a single computer. Hiding all the distribution from the users as well as the application programs can be achieved at two levels: 1) hide the distribution from users 2) at a lower level, make the system look transparent to programs. 1) and 2) requires uniform interfaces such as access to files, communication

33 65 In flexibility there are two schools of thoughts One school maintains that each machine should run a traditional kernel that provides most services itself. The other maintains that the kernel should provide as little as possible with the bulk of the operating system services available from user level servers. This two models known as the monolithic kernel and microkernel respectively

34 Make it easier to change Monolithic Kernel: systems calls are trapped and executed by the kernel. All system calls are served by the kernel, e.g., UNIX. Microkernel: provides minimal services. 1) IPC 2) some memory management 3) some low-level process management and scheduling 4) low-level i/o E.g., Mach can support multiple file systems, multiple system interfaces

35 Distributed system should be more reliable than single system. Example: 3 machines with 0.95 probability of being up. The probability of all being down will be Availability: fraction of time the system is usable. Redundancy improves it. Need to maintain consistency Need to be secure Fault tolerance: need to mask failures, recover from errors. 69 Without gain on this, why bother with distributed systems. Performance loss due to communication delays: fine-grained parallelism: high degree of interaction coarse-grained parallelism: data is communicated infrequently, after larger amounts of computation. Fault Tolerance also exacts its price

36 Systems grow with time or become obsolete. Techniques that require resources linearly in terms of the size of the system are not scalable. e.g., broadcast based query won't work for large distributed systems. Examples of bottlenecks Centralized components: a single mail server Centralized tables: a single URL address book Centralized algorithms: routing based on complete information

37

38 Omega network: 2x2 switches for n CPUs and n memories, log 2 n switching stages, each with n/2 switches, total (n log 2 n)/2 switches

39 If n=1024, How many switched stages will be there? How many will be the total switches stages (CPU to Memory & Memory CPU)? If a RISC processor is running at 100MIPS in then, How much would be the instruction execution time? 77 A multi-computer with 256-CPUs is organized as a 16X16 grid. What is the worst-case delay (in hops) that a message might have to take? 78 39

40 Consider a 256 CPU-hypercube what is the worst case delay here again in hops? 79 A multiprocessor has 4096 CPUs connected to memory by an omega network. Find out how many switches stages required? How many total switches stages required? How many total switches required? 80 40

41 A multiprocessor has MIPS CPUs connected to memory by an omega network. How fast do the switches have to be to allow a request to go to memory and back in one instruction time? 81 1.Explain the following : a) What is fine grained parallelism coarse grained parallelism? -2M b) Advantages of distributed systems over independent PCs and centralized systems- 2M 2.Discuss software concepts of distributed system. -6M 3.Explain design issues of distributed systems -6M (CO-1) 82 41