1 / 54

Introduction

Introduction. CSE 5306 Lecture Quiz due 3 June 2014. 1. History. 1980s: the Altair 1990s Novell Netware 2000s Distributed Systems. R U O K ?. 1. Which of the following was NOT a precursor of today’s distributed system? a. Low -cost, high-performance minicomputers.

kareem
Download Presentation

Introduction

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Introduction CSE 5306 Lecture Quiz due 3 June 2014

  2. 1. History • 1980s: the Altair • 1990s Novell Netware • 2000s Distributed Systems

  3. R U O K ? 1. Which of the following was NOT a precursor of today’s distributed system? a. Low-cost, high-performance minicomputers. b. High-speed local-area networks. c. Mobile phones.

  4. 1.1 Definition • “A distributed system is a collection of independent computers that appears to its user as a single coherent machine.” • Examples: • Many LAN processors sharing a common middleware. • Embedded processors in automotive or avionics systems. • Distant Early Warning (DEW) system’s many radars.

  5. R U O K ? 2. How does Tanenbaum define “distributed system”? a. Four jet fighters synergistically sharing target recognition radars. b. Collaborating autonomous computers that appear to be a coherent system. c. Enabling a whole new collection of desirable user experiences.

  6. R U O K ? 3. What purpose does middleware serve in a distributed system? a. It joins disparate computer hardware and software. b. It standardizes computer interfaces. c. It hides computers’ differences from the user. d. All of the above. e. None of the above.

  7. 1.2 Distributed System Goals Distributed systems…. • Make resources accessible. • Keep apps’ implementations transparent. • Open a system’s syntax & semantics. • Scale systems according to demand.

  8. R U O K ? 4. According to Tanenbaum, which of the following is NOT a distributed system design goal? a. Enhance the user’s experience. b. Make resources accessible. c. Keep app implementations transparent. d. Open the system’s syntax & semantics. e. Scale systems according to demand.

  9. 1.2.1 Making Resources Accessible • White board collaborations span great distances. • Sharing printers reduces costs. • Amazon’s mega warehouses are as close as your cell phone.

  10. Distributed System Complexity Mouse Manufactured to Military Specifications

  11. R U O K ? 5. A distributed system might makewhich of the following resources available? a. Decaffeinated coffee. b. Roses on Valentine’s Day. c. Costly, seldom used peripheral devices. d. All of the above. e. None of the above.

  12. R U O K ? 6. Which of the following problems might arise from designing a friendly user interface? a. Excessive system usage. b. Sabotage. c. Over budget development cost. d. All of the above. e. None of the above.

  13. 1.2.2 Distribution Transparency • A distributed system that is able to present itself to users and applications as if it were only a single computer system is said to be “transparent.”

  14. Access Transparency Your GPS screen …hides its (obscure” data representation and how it accesses its (very technical) resources.

  15. Location, Relocation & Migration Transparencies Location: People can call your cell phone (or email you) without knowing where you are. Migration: You can move your cell phone from your home to your office without callers knowing it. Relocation: You can travel from home to your office, while talking on your cell phone, without a caller knowing it.

  16. Replication Transparency • Replicating a New York Stock Exchange trading terminal inside Morgan-Stanley’s Dallas brokerage firm is a great convenience to the trader. • But it cannot be used for “algorithm trading,” which depends upon nanosecond updates. (Such computers are located only a few meters away from stock exchange databases.)

  17. Concurrency Transparency • Morgan-Stanley’s terminal must not be concurrently connected with another replicate in (competitor) Chase Bank’s brokerage firm. • While Morgan-Stanley is buying or selling a stock, Chase must be locked out of that stock’s mutual-exclusion code segment. • Chase should not even know that Morgan-Stanley is logged in.

  18. Failure Transparency • Twenty minutes ago, you heard the power transformer in your neighbor’s yard blow up. • A TXU service truck just pulled up. • But your electric oven is is still baking tonight’s chicken dinner. • That’s “failure transparency” (fault tolerance) hard at work in a very well designed (and expensive) distributed system. • A slow response that looks like a failure can wreak havoc on the grid.

  19. Undesired Transparency • Access to resources several thousand miles away should not pretend to have concurrent transparency. • A distributed system should not pretend that a failed resource is simply slow, by silently moving to another resource. • Location hiding should not cause my print job to be rerouted from my busy printer to an idle one in a distant city. • Access to resources several thousand miles away should not pretend to have concurrent transparency. • A distributed system should not pretend that a failed resource is simply slow, by silently moving to another resource. • Location hiding should not cause my print job to be rerouted from my busy printer to an idle one in a distant city. • [Bean person]

  20. 1.2.3 Openness • An Interface Definition Language (IDL) translates the user’s syntax into the native resources’ apps and their parameters. • The designer makes a distributed system “open” by crafting a natural language interface to embody the users’ semantics. • [car dashboard & car with hood open] • An Interface Control Document (ICD) contains detailed specifications of resource interfaces that are not open to users. • “Interoperable” resources are easily swapped. • “Portable” apps run well on a wide variety of systems. • Resources can be added easily to “extensible” systems.

  21. Scalability Distributed System # Users Administrative (Ease of Mgmt) Geographical • Scalability can be measured along three dimensions: • Total number of users. • Physical separation of machines. • Ease of management by a variety of organizations.

  22. Scalability Limitations • Characteristics of decentralized algorithms: • No machine has complete information about the system state. • Machines make decisions based only on local information. • Failure of one machine does not ruin the algorithm. • There is no implicit assumption that a global clock exists.

  23. Scaling Techniques • Domain Name Service divides a global name space into manageable zones. • Server distributes forms management to its clients’ apps.

  24. 1.2.5 Distributed System Pitfalls • Network reliability, security and homogeneity are design requirements unique to distributed systems. • Too often distributed system designers mistakenly assume: • Fixed topology. • Zero latency and infinite bandwidth. • Zero transport costs. • A single, competent administrator.

  25. Distributed Systems Family Distributed Systems Distributed Computing Systems Distributed Embedded Systems Distributed Information Systems Transaction Processing Systems Home Systems Sensor Networks Clusters Office Databases Grids Electronic Healthcare Systems Ubiquitous Computing Environments The various types of distributed systems can be sorted into the taxonomy above.

  26. Cluster • Thousands of identical parallel processors have taken over traditional supercomputers’ big math. • Beowulf’s master allocates batch jobs among slaves. • MOSIX’ processes transparently travel from home node to others for resources, appearing to be a single system.

  27. Grid • Grids are disparate collaborating groups that “virtually organize” to share expensive resources (e.g., CERN). Middle -ware SOA

  28. R U O K ? 7. Which of the following is an example of a distributed computing system? a. An information computing system. b. A cluster computing system. c. An embedded computing system. d. A grid computing system. e. Both b and d above.

  29. R U O K ? 8. How do clusters and grids differ? a. The workstations in a grid are all the same; clusters’ can differ widely. b. Clusters unite collaborators into a virtual organization; grid users are in one company. c. Clusters have one administrator; grids are federated. d. Middleware connects cluster apps and resources; two networks connect them in a grid. e. The SOA and SGSA standards specify cluster architectures; grids are ad hoc.

  30. R U O K ? 9. Beowulf… a. Is a grid computing system. b. Enables processes to transparently travel from their home nodes to other nodes for resources. c. Attempts to provide a single-system image. d. Consists of a master that allocates batch jobs among slave processors. e. All of the above.

  31. R U O K ? Match the following grid layers with their definitions below: 10. Resource __ 11. Fabric __ 12. Applications__ 13. Collective __ 14. Connectivity __ a. Configures interfaces to local resources with a cross-point switch. b. Authenticates users’ apps and local resources, then communicates by protocols with resources. c. Provides hardware driver DLLs. d. Discovers needed resources and schedules their access. e. Contains the users’ applications.

  32. R U O K ? 15. Which of the following is true? a. All grid layers at an administrative unit’s site are collectively called an open grid services architecture (OGSA). b. The service-oriented architecture (SOA) is a standard that specifies the OGSA, plus all of its component parts. c. All but the top and bottom layers of a grid computing system are called “middleware.” d. MOSIX is one example of a grid computing system. e. The three major types of distributed systems are distributed computing systems, distributed information systems and distributed real-time systems.

  33. Distributed Information Systems • Distributed data base systems dominate traditional offices. • The enterprise application integration (EIA) industry standardizes multiple user’s multiple database service transactions. • Transaction processing systems move money securely in banks.

  34. Transaction Processing Systems • Every uninterruptable, all-or-nothing transaction has a BEGIN and END. • A transaction starts with a private copy of its bank’s ledger, which it changes and writes back to the bank after it commits. • Transactions are ACID: • Atomic (indivisible) • Consistent (obeys banks’ Law of Conservation of Money) • Isolated (parallel tasks appear serial) • Durable (changes are permanent.

  35. Transaction Processing Systems • Transactions also can be nested (fork & join), but sub-transactions are not durable. • Remote procedure calls (RPCs) direct distant servers to withdraw and direct money, send and receive mail, etc. • A transaction processing monitor (TPM) could handle three different airline reservation servers (above), for travelers planning a trip.

  36. Enterprise Application Integration • Today’s transaction processing apps autonomously talk with each other (no TPM intervention) via… • middleware (above). • RPCs (remote procedure calls for apps). • RMIs (remote method invocations of objects). • Publish/subscribe message-oriented middleware (MOM) handles tightly coupled app communications, when the subscriber might not always be available.

  37. R U O K ? 16. Automated banking transactions must be… a. Atomic; i.e., indivisible. b. Consistent; e.g., obedient to the Law of Conservation of Money. c. Isolated; i.e., parallel tasks appear to happen serially. d. Durable; i.e., changes are permanent. e. All of the above.

  38. R U O K ? Match the following abbreviations with their definitions below: 17. EIA __ 18. TPM__ 19. RPC__ 20. RMI __ 21. MOM__ a. Publish/subscribe middleware. b. Execute distant servers’ apps. c. Standardizes many user’s multiple database service transactions. d. Execute distant servers’ methods. e. CoCoordinate transactions on many servers.

  39. R U O K ? 22. Which of the following describe(s) a “transaction”? a. Contacts distant servers to withdraw and direct money, send and receive mail, etc. b. Uninterruptable, all-or-nothing. c. Its parts may not be durable. d. Makes private copy of bank ledger, changes it and replaces the original with it. e. All of the above.

  40. R U O K ? 23. In the Enterprise Application Integration paradigm, transaction processing apps can talk with each other via middleware. a. True. b. False.

  41. R U O K ? 24. Publish/subscribe middleware 1) accepts a message from any app at any time and 2) delivers an app’s designated messages whenever the app is available to receive them. a. True. b. False.

  42. Distributed Pervasive Systems • WIFI laptops and mobile phones are designed to… • To adapt to new environments without administrator intervention • To embrace context changes; e.g., cell tower hand-offs • To enable user preferences • To share data freely

  43. Home Systems • Universal plug-and play standards (e.g., Bluetooth) join… • personal computers • TVs • video games • security systems • personal wearable electronics • Centralizing stored data saves money on low-capacity cell phones. • Privacy concerns: • Can the FBI seize (4th Amendment rights) the personal data you store in the cloud? Yes, it is outside the home in a public place. • Privacy-concerned small towns are selling UAV Hunting Licenses. • “Recommenders” automatically organize your personal data like they see others doing it.

  44. Electronic Healthcare Systems • Preventive care keeps people out of expensive hospitals. • A wireless body area network (BAN) may… • May autonomously gather data to periodically upload from its hub • May be controlled by a physician’s external network (above). • Design considerations: • Where to store data? • how to prevent its loss? • what alerts to generate? • when to intervene? • how reliable is it? • how secure is it?

  45. Sensor Networks • Monitoring bridge safety, traffic, weather and volcanoes call for thousands of sensors organized as a distributed database (above). • They may accept queries or spontaneously deliver data at regular intervals or upon an event. • Aggregating data with TinyDB’s in-network data processing can dramatically reduce sensor tree-network traffic. • Publish/subscribe middleware may optimize data transfers among temporally and geographically dispersed sensors.

  46. R U O K ? 25. Privacy concerns in distributed pervasive systems include: a. Amazon’s UAV (unpiloted air vehicle) possibly glancing into bedroom windows, while delivering books. b. Google looking into a great many other users’ personal data, so that it can organize your data in the usual way. c. The NSA eavesdropping on your phone conversations, while intending to gather only “metadata.” d. Traffic surveillance cameras identifying you by your Facebook photograph, so as to track your suspicious movements late at night. e. All of the above.

  47. R U O K ? 26. Of what legal liability problems should an electronic healthcare system designer be wary? a. Accidental loss of a patient’s Holter heart monitor data that includes a life-threatening arrhythmia. b. Hackers turning off a controversial politician’s web-enabled cardiac pacemaker. c. Propagation of timely and accurate of alerts to health monitoring personnel. d. Doctors’ prescribing FDA-certified cell phone apps that help diabetics control their blood sugar. e. All of the above.

  48. R U O K ? 27. Which of the following “experience enhancing” features have users come to expect in today’s wireless devices? a. Ability to communicate with all Bluetooth enabled devices in the local area. b. Embracing contextual changes, such as a distant cell tower handing off a mobile phone to a closer one. c. Cell phone automatically switching to your car’s speaker and microphone, when you step inside the vehicle. d. Automatically uploading photos to the cloud, when your cheap cell phone overflows. e. All of the above.

  49. R U O K ? 28. Which of the following should be considered in the design of 10,000 networked bridge sensors? a. Any sensor should be able to spontaneously deliver data that suggests imminent structural failure. b. The sensors should be organized in a tree network, responding to queries like a tiny database. c. Temporally and geographically dispersed sensors may communicate via publish/subscribe middleware. d. All of the above. e. None of the above.

  50. Summary • See pp.30, 31 in your textbook.

More Related