Introduction Chapter 1 Kyung Hee University 1 41

Introduction Chapter 1 Kyung Hee University 1/41

Agenda -Definition -Hardware concept -Software concept -The Client-Server model Kyung Hee University 2/41

Definition of a Distributed System (1) A distributed system is: A collection of independent computers that appears to its users as a single coherent system. Kyung Hee University 3/41

Definition of a Distributed System (2) 1. 1 A distributed system organized as middleware. Note that the middleware layer extends over multiple machines. Kyung Hee University 4/41

Goal of Distributed system • • Connecting users to resources Openness Transparency Scalability Kyung Hee University 5/41

Transparency in a Distributed System Transparency Description Access Hide differences in data representation and how a resource is accessed Location Hide where a resource is located Migration Hide that a resource may move to another location Relocation Hide that a resource may be moved to another location while in use Replication Hide that a resource is replicated Concurrency Hide that a resource may be shared by several competitive users Failure Hide the failure and recovery of a resource Persistence Hide whether a (software) resource is in memory or on disk Different forms of transparency in a distributed system. Kyung Hee University 6/41

Scalability Problems Concept Example Centralized services A single server for all users Centralized data A single on-line telephone book Centralized algorithms Doing routing based on complete information Examples of scalability limitations. Kyung Hee University 7/41

Scaling Techniques (1) 1. 4 The difference between letting: a) a server or b) a client check forms as they are being filled Kyung Hee University 8/41

Scaling Techniques (2) 1. 5 An example of dividing the DNS name space into zones. Kyung Hee University 9/41

Hardware Concepts 1. 6 Different basic organizations and memories in distributed computer systems Kyung Hee University 10/41

Multiprocessors (1) 1. 7 A bus-based multiprocessor. Kyung Hee University 11/41

Multiprocessors (2) 1. 8 Kyung Hee University a) A crossbar switch b) An omega switching network 12/41

Multicomputers • Each node is an autonomous machine q. Private memory • Lower traffic than multiprocessors q. CPU – CPU versus CPU – Memory traffic • Homogeneous or Heterogeneous Kyung Hee University 13/41

Homogeneous Multicomputer Systems (1) • Similar nodes q. Same processors and memory space • Homogeneous access to network q. Single network • Bus-based or point-to-point communication Kyung Hee University 14/41

Homogeneous Multicomputer Systems (2) 1 -9 Kyung Hee University a) Grid b) Hypercube 15/41

Heterogeneous Multicomputers • Different nodes q. Nodes can be complex system • None-homogeneous access to network q. Different network • Distributed systems are commonly built on this H/W category q. Need S/W to make it transparent Kyung Hee University 16/41

Software Concepts System Description Main Goal DOS Tightly-coupled operating system for multiprocessors and homogeneous multicomputers Hide and manage hardware resources NOS Loosely-coupled operating system for heterogeneous multicomputers (LAN and WAN) Offer local services to remote clients Middleware Additional layer atop of NOS implementing general-purpose services Provide distribution transparency An overview between • DOS (Distributed Operating Systems) • NOS (Network Operating Systems) • Middleware Kyung Hee University 17/41

Uniprocessor Operating Systems 1. 11 Kyung Hee University Separating applications from operating system code through a microkernel. 18/41

Multicomputer Operating Systems (1) 1. 14 General structure of a multicomputer operating system Kyung Hee University 19/41

Multicomputer Operating Systems (2) 1. 15 Alternatives for blocking and buffering in message passing. Kyung Hee University 20/41

Distributed Shared Memory Systems (1) a) b) c) Pages of address space distributed among four machines Situation after CPU 1 references page 10 Situation if page 10 is read only and replication is used Kyung Hee University 21/41

Distributed Shared Memory Systems (2) 1. 18 False sharing of a page between two independent processes. Kyung Hee University 22/41

Network Operating System (1) 1 -19 General structure of a network operating system. Kyung Hee University 23/41

Network Operating System (2) 1 -20 Two clients and a server in a network operating system. Kyung Hee University 24/41

Network Operating System (3) 1. 21 Different clients may mount the servers in different places. Kyung Hee University 25/41

DOS and NOS v. s DS • DOS qualifies as DS? q. Computers are not independent q. Easy to use and transparent • NOS qualifies as DS? q. No single coherent view q. Scalable and open Kyung Hee University 26/41

Positioning Middleware 1 -22 General structure of a distributed system as middleware. Kyung Hee University 27/41

Middleware and Openness 1. 23 In an open middleware-based distributed system, the protocols used by each middleware layer should be the same, as well as the interfaces they offer to applications. Kyung Hee University 28/41

Comparison between Systems Item Distributed OS Network OS Middlewarebased OS Multiproc. Multicomp. Very High Low High Yes No No Number of copies of OS 1 N N N Basis for communication Shared memory Messages Files Model specific Resource management Global, central Global, distributed Per node Scalability No Moderately Yes Varies Openness Closed Open Degree of transparency Same OS on all nodes A comparison between multiprocessor operating systems, multicomputer operating systems, network operating systems, and middleware based distributed systems. Kyung Hee University 29/41

The Clients and Servers Model Kyung Hee University 30/41

Client and Server -Clients request services. -Servers provide services by replying to the requests. General interaction between a client and a server. Kyung Hee University 31/41

An Example Client and Server (1) The header. h file used by the client and server. Client and server need to agree on a couple things first! Kyung Hee University 32/41

An Example Client and Server (2) A sample server (A basic server loop) -The server continually listens for new requests (receive). -Extracting the request and simply executes it. -It then prepares a response message and sends it back to the requesting client. Kyung Hee University 33/41

An Example Client and Server (3) Sample Client: using the server to copy a file. 1 -27 b Kyung Hee University 34/41

Application Layering A client-server application typically consists of three layers: 1. User-Interface level: +Consists of the programs that allow end users to interact with application. 2. Processing level: +Implements the application logic (core functionality) +Typically implemented at the server side 3. Data level +Maintain the actual data on which the application operate. +Also keep data consistent across different application Kyung Hee University 35/41

Application Layering Web browser 1 -28 Application server Database server The general organization of an Internet search engine into three different layers Kyung Hee University 36/41

Client-Server Architectures Two-tiers Architectures 1 -29 Alternative client-server organizations (a) – (e). Kyung Hee University 37/41

Client-Server Architectures (con. t) Three-tiers Architectures 1 -30 An example of a server acting as a client. Kyung Hee University 38/41

Modern Architectures -Vertical distribution Model : where the tiers corresponded directly with the logical organization of the apps (The previous architectures) -Horizontal distribution Model: where the application layers are physically split up into logically equivalent part -> This is how high-performance web services (e. g. , Google, Amazon, E-bay. . ) An example of horizontal distribution of a Web service. Kyung Hee University 39/41

End of Chapter 1 Kyung Hee University 40/41

Communication Chapter 2 Kyung Hee University 41/63

Agenda 1 -Layered protocols 2 -Remote Procedure 3 -Remote Object Invocation 4 -Message-Oriented Communication 5 -Stream-Oriented Communication Kyung Hee University 42/63

1 - Layered Protocols Kyung Hee University 43/63

OSI Model -OSI: Open Systems Interconnection -Developed by the International Organization for Standardization (ISO) -Provides a generic framework to discuss the layers and functionalities of communication protocols. Kyung Hee University Layers, interfaces, and protocols in the OSI model. 44/63

OSI Model (con. t) 2 -2 A typical message as it appears on the network. Kyung Hee University 45/63

OSI Protocol Layers -Physical layer +Deals with the transmission of bits +Physical interface between data transmission device (e. g. computer) and transmission medium or network Concerned with: • Characteristics of transmission medium, Signal levels, Data rates -Data link layer: +Deals with detecting and correcting bit transmission errors +Bits are group into frames +Checksums are used for integrity Kyung Hee University 46/63

OSI Protocol Layers (con. t) Discussion between a receiver and a sender in the data link layer. Kyung Hee University 47/63

OSI Protocol Layers (con. t) -Network layer: +Performs multi-hop routing across multiple networks +Implemented in end systems and routers -Transport layer: +Packing of data +Reliable delivery of data (breaks message into pieces small enough, assign each one a sequence number and then send them) +Ordering of delivery Examples: • TCP (connection-oriented) • UDP (connectionless) • RTP (Real-time Transport Protocol) Kyung Hee University 48/63

OSI Protocol Layers (con. t) Client-Server TCP protocol (a) Normal operation of TCP. (b) Transactional TCP. Kyung Hee University 49/63

OSI Protocol Layers (con. t) -Session layer +Provide dialog control to keep track of which party is talking and it provides synchronization facilities -Presentation layer +Deals with non-uniform data representation and with compression and encryption -Application layer +Support for user applications e. g. HTTP, SMPT, FTP Kyung Hee University 50/63

Middleware Protocols -Support high-level communication services -The session and presentation layers are merged into the middleware layer, Ex: Microsoft ODBC (Open Database Connectivity), OLE DB… An adapted reference model for networked communication. Kyung Hee University 51/63

2 - Remote Procedure Call Kyung Hee University 52/63

Remote Procedure call -Basic idea: To execute a procedure at a remote site and ship the results back. -Goal: To make this operation as distribution transparent as possible (i. e. , the remote procedure call should look like a local one to the calling procedure). Example: read(fd, buf, nbytes) Kyung Hee University 53/63

Client and Server Stubs Definition: Are additional functions which are added to the main functions in order to support for RPC Client count = do. Something(); Client Stub OS Kyung Hee University Server procedure do. Something(); Server Stub OS 54/63

Steps of a Remote Procedure Call 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Client procedure calls client stub in normal way Client stub builds message, calls local OS Client's OS sends message to remote OS Remote OS gives message to server stub Server stub unpacks parameters, calls server Server does work, returns result to the stub Server stub packs it in message, calls local OS Server's OS sends message to client's OS Client's OS gives message to client stub Stub unpacks result, returns to client Kyung Hee University 55/63

Passing Value Parameters (1) 2 -8 Steps involved in doing remote computation through RPC Kyung Hee University 56/63

Passing Value Parameters (2) In a large distributed system, multiple machine types are present Each machine has its own representation for number, characters, and others data items. a) b) c) Original message on the Pentium (little-endian) The message after receipt on the SPARC (big-endian ) The message after being inverted. The little numbers in boxes indicate the address of each byte Kyung Hee University 57/63

Parameter Specification -Caller and callee agree on the format of message they exchange Ex: word = 4 bytes float = 1 word character is the rightmost byte of word => the client stub must use this format and the server stub know that incoming message for foobar has this format Kyung Hee University 58/63

Asynchronous RPC (1) -Avoids blocking of the client process. -Allows the client to proceed without getting the final result of the call. 2 -12 a) b) Kyung Hee University The interconnection between client and server in a traditional RPC The interaction using asynchronous RPC 59/63

Deferred Synchronous RPC(2) One-way RPC model: client does not wait for an acknowledgement of the server’s acceptance of the request. 2 -13 A client and server interacting through two asynchronous RPCs Kyung Hee University 60/63

Example DCE RPC -What is DCE ? (Distributed Computing Environment) DCE is a true middleware system in that it is designed to execute as a layer of abstraction between exiting (network) operating system and distributed application. -Goals of DCE RPC Makes it possible for client to access a remote service by simply calling a local procedure. -Components: +Languages +Libraries +Daemon +Utility programs +Others Kyung Hee University 61/63

Writing a Client and a Server Generate a prototype IDL file containing an interface identify Filling in the names of remote procedure and their parameters IDL compiler is call to compile into three files Kyung Hee University 2 -14 The steps in writing a client and a server in DCE RPC. 62/63

Binding a Client to a Server 2 -15 Endpoint (port) is used by server’s OS to distinguish incoming message Client-to-server binding in DCE. Kyung Hee University 63/63

3 -Remote Object Invocation Kyung Hee University 64/63

Distributed Objects (1) • Objects separate their actual implementation from their interface • Distributed object = an object which publishes its interface on other machines • Remote object = a distributed object whose state is centralized Kyung Hee University 65/63

Distributed Objects (2) 2 -16 Common organization of a remote object with client-side proxy. Kyung Hee University 66/63

Binding a Client to an Object (1) When a client binds to a distributed object, an implementation of the object interface (proxy) is loaded into the client’s address space Kyung Hee University 67/63

Binding a Client to an Object (2) Distr_object* obj_ref; obj_ref = …; obj_ref-> do_something(); //Declare a systemwide object reference // Initialize the reference to a distributed object // Implicitly bind and invoke a method (a) Distr_object obj. Pref; Local_object* obj_ptr; obj_ref = …; obj_ptr = bind(obj_ref); obj_ptr -> do_something(); //Declare a systemwide object reference //Declare a pointer to local objects //Initialize the reference to a distributed object //Explicitly bind and obtain a pointer to the local proxy //Invoke a method on the local proxy (b) a) An example with implicit binding using only global references b) An example with explicit binding using global and local references Problems: + Language dependent + Address of server an Object reference Kyung Hee University 68/63

Parameter Passing (2) • Pass remote object by reference • Pass local object by value • Local object = an object in the client’s address space Kyung Hee University 69/63

Parameter Passing (2) 2 -18 The situation when passing an object by reference or by value. Kyung Hee University stream 70/63

Example: Java RMI (1) Java Distributed-Object Model • Distributed objects integrated into the Language • Goal is to achieve transparency! (keep as much of semantics of nondistributed objects as possible) Kyung Hee University 71/63

Java RMI (2) Local vs. Remote object differences when 1. Cloning: Cloning the actual object only, and not its proxies 2. Synchronizing • Synchronized methods • Only proxy synchronization is allowed only Kyung Hee University 72/63

Java RMI (3) ØAny serializable object type can be a parameter to an RMI Platform dependent objects (e. g. , sockets, file descriptors, etc) are not serializable ØLocal objects are passed by value, remote objects are passed by reference ØProxy can be used as a reference to a remote object: Possible to serialize the proxy and send it to another server Hee Kyung 73/63 University

Java RMI (4) Kyung Hee University stream 74/63

4 -Message-Oriented Communication Kyung Hee University 75/63

Persistence and Synchronicity in Communication General organization of a communication system in which hosts are connected through a network +Each host is connected to a single communication server. +Hosts and communication servers can have buffers. Kyung Hee University 76/63

Persistence and Synchronicity in Communication Persistent communication of letters back in the days of the Pony Express Kyung Hee University 77/63

Persistence and Synchronicity in Communication Definition -Persistent vs. Transient q Persistent messages are stored as long as necessary by the communication system (e. g. , e-mail) q Transient messages are discarded when they cannot be delivered (e. g. , TCP/IP) -Synchronous vs. Asynchronous q Asynchronous implies sender proceeds as soon as it sends the message no blocking q Synchronous implies sender blocks till the receiving host buffers the message Kyung Hee University 78/63

Persistence and Synchronicity in Communication (a) Persistent asynchronous communication (b) Persistent synchronous communication Kyung Hee University 79/63

Persistence and Synchronicity in Communication (c) Transient asynchronous communication (d) Receipt-based transient synchronous communication Kyung Hee University 80/63

Persistence and Synchronicity in Communication (e) Delivery-based transient synchronous communication at message delivery (f) Response-based transient synchronous communication Kyung Hee University 81/63

Message-Orient Transient Communication q Berkeley Sockets: q Socket: a communication endpoint to which an application data (be sent to network) and read incoming data. Primitive Create a new communication endpoint Bind Attach a local address to a socket Listen Announce willingness to accept connections Accept Block caller until a connection request arrives Connect Actively attempt to establish a connection Send some data over the connection Receive some data over the connection Close Kyung Hee University Meaning Socket can write Release the connection Socket primitives for TCP/IP 82/63

Message-Orient Transient Communication q Berkeley Sockets: Create a new endpoint Associate endpoint Reserve buffer Block waiting for reqs Automatic binding after connection Connection-oriented communication pattern using sockets Kyung Hee University 83/63

Message-Orient Transient Communication q. Sockets ware deemed insufficient because: q. Support only send and receive primitives q. Designed for communication using general-purpose protocol stacks such as TCP/IP q->The Message-Passing Interface (MPI) q. Designed for multiprocessor machines and high-performance parallel programming q. Provides a high-level of abstraction than sockets q. Support diverse forms of buffering and synchronization (over 100 functions) Kyung Hee University 84/63

Message-Orient Transient Communication q The Message-Passing Interface (MPI) Primitive Meaning MPI_bsend Append outgoing message to a local send buffer MPI_send Send a message and wait until copied to local or remote buffer MPI_ssend Send a message and wait until receipt starts MPI_sendrecv Send a message and wait for reply MPI_isend Pass reference to outgoing message, and continue MPI_issend Pass reference to outgoing message, and wait until receipt starts MPI_recv Receive a message; block if there are none MPI_irecv Check if there is an incoming message, but do not block Some of the most intuitive message-passing primitives of MPI. Kyung Hee University 85/63

Message-Orient Persistent Communication q Message-Queue Model q Apps communicate by inserting messages in specific queues q Loosely-couple communication q Support for: q Persistent asynchronous communication q Longer message transfers • (e. g. , e-mail systems) Four combinations for loosely-coupled communications using queues. Kyung Hee University 86/63

Message-Orient Persistent Communication q General Architecture of a Message-Queuing System q Messages can only be put and received from local queues. q The message-queuing system is responsible for transmitting the messages between the source queues and destination queues, meanwhile storing the messages as long as necessary. q Each queue is maintained by a queue manager. Example The relationship between queue-level addressing and network-level addressing. Kyung Hee University 87/63

Message-Orient Persistent Communication q General Architecture of a Message-Queuing System q Queue managers are not only responsible for directly interacting with applications but are also responsible for acting as relays (or routers). Queue managers form an overlay network, acting as routers Kyung Hee University 88/63

Message-Orient Persistent Communication q General–purpose of a Message-Queuing System q Enable persistent communication between processes q Handling access to database q Perform computation q… In wide range of application, include: -Email -Workflow -Groupware -Batch processing Kyung Hee University 89/63

Message-Orient Persistent Communication q Message Brokers The general organization of a message broker in a message-queuing system Kyung Hee University end 90/63

5 -Stream-Oriented Communication Kyung Hee University 91/63

Kyung Hee University 92/63

* Relationship between substreams are also timedependent Kyung Hee University 93/63

Data Stream (1) Setting up a stream between two processes across a network. Setting up a stream directly between two devices. Kyung Hee University 94/63

Data Stream (2) An example of multicasting a stream to several receivers. Kyung Hee University 95/63

Client Traffic Shaping Kyung Hee University 96/63

Specifying Qo. S (1) Characteristics of the Input • maximum data unit size (bytes) • Token bucket rate (bytes/sec) • Toke bucket size (bytes) • Maximum transmission rate (bytes/sec) Service Required • Loss sensitivity (bytes) • Loss interval ( sec) • Burst loss sensitivity (data units) • Minimum delay noticed ( sec) • Maximum delay variation ( sec) • Quality of guarantee A flow specification. Question: who specifies the flow? Kyung Hee University 97/63

Specifying Qo. S (2) The principle of a token bucket algorithm. Kyung Hee University 98/63

Resource Re. Ser. Vation Protocol (RSVP) A transport-level control protocol • Used to provide Qo. S for continuous data streams by reserving resources {Bandwidth, buffers, and processing capacity} • Issue: how to translate Qo. S parameters to resource usage? 2 ways to translate • RSVP translates Qo. S parameters into data link layer parameters • Data link layer provides its own set of parameters (as in ATM) Kyung Hee University 99/63

Setting Up a Stream (1) The basic organization of RSVP for resource reservation in a distributed system. Kyung Hee University 100/63

Setting Up a Stream (2) Kyung Hee University 101/63

Setting Up a Stream (3) Kyung Hee University 102/63

Setting Up a Stream (4) Kyung Hee University 103/63

Kyung Hee University 104/63

Synchronization Mechanisms (1) The principle of explicit synchronization on the level data units. Kyung Hee University 105/63

Synchronization Mechanisms (2) 2 -41 The principle of synchronization as supported by high-level interfaces. Kyung Hee University message 106/63

END OF CHAPTER 2 Kyung Hee University 107/63