Distributed Systems CS 3850 Soufiane Noureddine Lectures MWF 14:00 – 14:50 (PE207D) Office Hours MW 11:00 – 12:00 (C520)

Introduction Chapter 1

Local Area Network: LAN Computer Networks A collection of computers communicating through an underlying network is called a computer network (in contrast to a single-processor system) Local Area Network: LAN 10 to 1000 Mb/sec Wide Area Network: WAN 64 kbps to gigabits/sec

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. A. Tanenbaum

Definition of a Distributed System (2) “You know you have one when the crash of a computer you never heard of stops you from getting any work done.” L. Lamport

Organization of a Distributed System 1.1 A distributed system organized as middleware. Note that the middleware layer extends over multiple machines.

Characteristics of a distributed System  Differences between computers are hidden  Communication between computers is hidden  Internal organization of the system is hidden  Single system image: interaction with the system is independent from location (and time)  Ease of extension  High availability

Examples of distributed Systems Network of workstations + Pool of processors: - Single file system (uniform naming scheme) - Processors are allocated dynamically when needed (load sharing) - System acts like a single-processor system 2. Workflow systems - Orders arrive dynamically (e.g. via laptops, cellular phones) - System assigns orders to the corresponding departments and initiates the needed business processes and users are unaware of the internal flow of orders - System acts like a centralized database WWW - No need to know where documents are stored (at least in theory) - Accessing remote documents is like accessing local ones

Goals of distributed Systems Connecting users and resources:  consequence: communication and collaboration (e.g. joint editing)  issue: security 2. Openness - Services should obey to standard rules specifying syntax and semantics - Services are specified in general as interfaces in an interface definition language - IDL includes rather syntax and no semantics - Specification of interface should be completed and neutral (w.r.t. implementation) - Openness promotes interoperability and portability Transparency (see next slides) Scalability (see next slides)

Transparency in a Distributed System 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. Degree of transparency: More transparency means less performance In reality: Systems are transparent only to certain degree

Network (e.g. Telecom) Clients Software Developers e.g. Company Network Provider

Network (e.g. Telecom) Clients Software Developers e.g. Company Company A Company B Network Provider

Network (e.g. Telecom) Clients Software Developers e.g. Company Company A Company B Network Provider

Network (e.g. Telecom) Clients Software Developers e.g. Company Company A Company B Network Provider

Network (e.g. Telecom) Clients Software Developers e.g. Company Company A Company B Network Provider

Network (e.g. Telecom) Clients Software Developers e.g. Company Applications Operating System & Hardware Network Provider

Network (e.g. Telecom) Clients Software Developers e.g. Company Applications AEM: Availability Enhancing Middleware Operating System & Hardware Network Provider

Scalability Problems A system is scalable when it is easy to extend without loss of performance. Concept Example Centralized services A single server for all users (e.g. a single DNS server) Centralized data A single on-line telephone book (e.g. single table in a frequently used database) Centralized algorithms Doing routing based on complete information Examples of scalability limitations. Centralized solutions tend to be non-scalable Decentralized solutions promote scalability

Decentralized Algorithms No machine has complete information about the system state Machines make decisions based only on local information Failures of one machine does not ruin the whole algorithm No assumption on the existence of a global time

Scaling Techniques (1) Ways to solve scalability problems: Hide communication latencies (geographical scalability) Use of distribution (in order to avoid bottlenecks) Use of replication Ad a) Batch processing: Asynchronous communication instead of synchronous communication Interactive applications: Reduce overall communication Example: Move part of computation from server to client (e.g. Java applet)

Scaling Techniques (2) 1.4 The difference between letting: a server or a client check forms as they are being filled

Scaling Techniques (3) Ad b) Use of distribution 1.5 An example of dividing the DNS name space into zones. Clientserver Z1: nl.vu.cs.fluit Server Z1  Client: address of Z2 Clientserver Z2: vu.cs.fluit …

Scaling Techniques (4) Ad c) Use of replication Replication raises: Availability: in general the primary goal Performance: By balancing the load among replicas By locating replicas close to users/clients Example: Caching (e.g. browser cache for used WWW pages) Main issue in connection with replication: consistency

Different basic organizations and memories in distributed computer Hardware Concepts 1.6 Different basic organizations and memories in distributed computer systems

A bus-based multiprocessor. Multiprocessors (1) 1.7 A bus-based multiprocessor. Issues: Scalability Cache consistency

Multiprocessors (2) A crossbar switch: n2 switches! 1.8 A crossbar switch: n2 switches! An omega switching network: less switches

Homogeneous Multicomputer Systems 1-9 Grid Hypercube

Software Concepts An overview of DOS (Distributed Operating Systems) Description Main Goal DOS Tightly-coupled operating system for multi-processors 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 of DOS (Distributed Operating Systems) NOS (Network Operating Systems) Middleware

Uniprocessor Operating Systems 1.11 Separating applications from operating system code through a microkernel.

Multiprocessor Operating Systems (1) monitor Counter { private: int count = 0; public: int value() { return count;} void incr () { count = count + 1;} void decr() { count = count – 1;} } A monitor to protect an integer against concurrent access.

Multiprocessor Operating Systems (2) monitor Counter { private: int count = 0; int blocked_procs = 0; condition unblocked; public: int value () { return count;} void incr () { if (blocked_procs == 0) count = count + 1; else signal (unblocked); } void decr() { if (count ==0) { blocked_procs = blocked_procs + 1; wait (unblocked); blocked_procs = blocked_procs – 1; } else count = count – 1; A monitor to protect an integer against concurrent access, but blocking a process.

Multicomputer Operating Systems (1) 1.14 General structure of a multicomputer operating system

Multicomputer Operating Systems (2) 1.15 Alternatives for blocking and buffering in message passing.

Synchronization point Reliable comm. guaranteed? Multicomputer Operating Systems (3) Synchronization point Send buffer Reliable comm. guaranteed? Block sender until buffer not full Yes Not necessary Block sender until message sent No Block sender until message received Necessary Block sender until message delivered Relation between blocking, buffering, and reliable communications.

Distributed Shared Memory Systems (1) 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 Replicating all pages:  Coherence protocols: a) strong (transparent) b) weak (not transparent)

Distributed Shared Memory Systems (2) Page size: small  more communication overhead large  less communication, but false sharing may occur 1.18 False sharing of a page between two independent processes. Problem with DSM: not as efficient as expected

Network Operating System - NOS (1) 1-19 General structure of a network operating system. Main features: Independent operating systems Services for accessing remote resources

Network Operating System (2) 1-20 Two clients and a server in a network operating system.

Network Operating System (3) 1.21 Different clients may mount the servers in different places.

NOS vs DOS (3) Distributed operating system: Fully transparent For homogeneous systems  computers are not independent More secure, but less scalable and less open Network operating system: Not transparent For heterogeneous system  computers are independent Easier to extend (e.g. adding a new node in the Internet) Both do not really qualify as a distributed system!!!  Solution: Middleware atop of a NOS hiding heterogeneity and improving transparency

Positioning Middleware 1-22 General structure of a distributed system as middleware. Middleware Services: rather a functionally complete set of services in order to hide heterogeneity, direct access to NOS is discouraged.

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.  implementations of the middleware should not use the NOS-provided protocols

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

General interaction between a client and a server. Clients and Servers 1.25 General interaction between a client and a server.

An Example Client and Server (1) The header.h file used by the client and server.

An Example Client and Server (2) A sample server.

An Example Client and Server (3) 1-27 b A client using the server to copy a file.

Processing Level 1-28 The general organization of an Internet search engine into three different layers

Multitiered Architectures (1) 1-29 Alternative client-server organizations (a) – (e).

Multitiered Architectures (2) 1-30 An example of a server acting as a client.

An example of horizontal distribution of a Web service. Modern Architectures 1-31 An example of horizontal distribution of a Web service.