School of Computing Science Simon Fraser University - PowerPoint PPT Presentation

arne
school of computing science simon fraser university n.
Skip this Video
Loading SlideShow in 5 Seconds..
School of Computing Science Simon Fraser University PowerPoint Presentation
Download Presentation
School of Computing Science Simon Fraser University

play fullscreen
1 / 39
Download Presentation
School of Computing Science Simon Fraser University
160 Views
Download Presentation

School of Computing Science Simon Fraser University

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. School of Computing ScienceSimon Fraser University CMPT 880: Large-scale Multimedia Systems and Cloud Computing Introduction Mohamed Hefeeda

  2. Course Objectives • Understand basics of multimedia systems & cloud computing • Know current research issues in these areas • Develop research skills • Reading papers, presentation skills, research discussion, finding project ideas, code development, and writing

  3. Course Info • Course web page http://nsl.cs.sfu.ca/teaching/13/880/ • References • Mostly research papers and book chapters

  4. Course Info: Grading • Class participation and Assignments: 50% • Present one topic, from chapter(s)/paper(s) • Read all Mandatory Reading and participate in discussion • Few assignments and quizzes • Final Project: 50% • New Research Idea (publishable  A+) • Implementation and evaluation of an already-published algorithm/technique/system (Good demo  A+) • Quantitative comparisons between two already-published algorithms/techniques/systems • A survey of a topic • …

  5. Course Info: Topics • Introduction • Overview of clouds and multimedia systems • Video coding basics • Cloud computing • Datacenter design • Virtualization • Storage systems • Programming models • Cloud support for multimedia systems • Mobile multimedia clouds

  6. Multimedia Systems

  7. Definitions and Motivations • “Multimedia” is an overused term • Means different things to different people • Because it touches many disciplines/industries • Computer Science/Engineering • Telecommunications Industry • TV and Radio Broadcasting Industry • Consumer Electronics Industry • …. • For users • Multimedia = multiple forms/representations of information (text, audio, video, …)

  8. Definitions and Motivations • Why should we study/research multimedia topics? • Huge interest and opportunities • High speed networks • Powerful (cheap) computers (desktops … cell phones) • Abundance of multimedia capturing devices (cameras, speakers, …) • Tremendous demand from users (mm content makes life easier, more productive, and more fun) • Here are some statistics …

  9. Some video statistics • Growth of various video traffic [Cisco 2008] • Video traffic accounted for 32% of Internet traffic in 2008 and is estimated to account for 50% in 2012 • Y-axis in Petabytes (1000 Terabytes) per month.

  10. Multimedia:The Big Picture [SN04]

  11. QoS in Networked Multimedia Systems • Quality of Service = “well-defined and controllable behavior of a system according to quantitatively measurable parameters” • There are multiple entities in networked multimedia system • User • Network • Local system (memory, processor, file system, …)

  12. QoS in Networked Multimedia Systems • Different parameters belong to different entities  QoS Layers

  13. QoS Layers Perceptual (window size, security) User Media Quality (frame rate, adaptation rules) Application System Network Local Devices Processing (CPU scheduling, memory, hard drive) Traffic (bit rate, loss, delay, jitter)

  14. QoS Layers • QoSSpecification Languages • Mostly application specific • XML based • See: Jin & Nahrstedt, QoS Specification Languages for Distributed Multimedia Applications: A Survey and Taxonomy, IEEE MultiMedia, 11(3), July 2004 • QoSmapping between layers • Map user requirements to Network and Device requirements • Some (but not all) aspects can be automated • For others, use profiles and rule-of-thumb experience • Several frameworks have been proposed in the literature • See: Nahrstedt et al., Distributed QoS Compilation and Runtime Instantiation, IWQoS 2000

  15. QoS Layers • QoSenforcement methods • The most important/challenging aspect • How do we make the network and local devices implement the QoS requirements of MM applications? • We need to • enforce QoS in Network (models/protocols) • enforce QoS in Processor (CPU scheduling for MM) •  When we combine them, we get end-to-end QoS • Notice: • If not enough resources, we have to adapt (or scale) the MM content (e.g., use smaller resolution, frame rate, drop a layer, etc)

  16. Cloud Computing

  17. Cloud Computing • “Cloud Computing” … fuzzy term • Some argue it is just rebranding of old stuff • Others see it as revolutionarytechnology that will transform everything in computing • Truth … somewhere in between

  18. Cloud Computing: Vision • Goal … achieve the old dream for computing Make computing a utility • Similar to electricity & water • we (customers) do not worry about design, operation, maintenance of power plants, nor do we think about power transmission systems • Home users  simple requirements, e.g., lighting • Industries  complex requirements, e.g., high voltage • … and we pay as we consume

  19. Cloud Computing: NIST Definition “Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.”

  20. Cloud Computing: Service Models • IaaS (Infrastructure as a Service) • Basic computing resources (CPU, storage, network, …) • Amazon EC2 • PaaS (Platform as a Service) • Platform to develop apps using programming languages, libraries, services, and tools supported by the cloud provider • Windows Azure, Amazon EMR (Elastic MapReduce) • SaaS (Software as a Service) • Software apps provided by the cloud provider • SalesForce.com (e.g., payroll, customer relation management, …)

  21. Cloud Computing: Why Now? • Better Internet & Mega Datacenters • Internet: faster, prevalent, and more reliable • Mega Datacenters: • economy of scale (5—7x cheaper hardware than medium size companies) • Already deployed (Amazon AWS, Google, …)  Additional revenue stream • Already developed software for in-house use (e.g., Google File System, MapReduce)

  22. Cloud Computing: Why Now? (2) • New technology trends and business models • Shifting from high-touch, -margin, -commitment to low-touch, -margin, -commitment service • E.g., content distribution using Akamai vs. using Amazon CloudFront • New application opportunities  • Mobile interactive apps, large batch processing, business analytics, …

  23. Cloud Computing: 3 New Aspects • Illusion of infinite computing resources  • Users do not need to plan ahead for provisioning • Elimination of up-front commitments by users  • Start small and increase on demand • Pay on short term basis, e.g., hourly  • Cost saving by getting machines only when needed • Elasticity: can scale up or down (quickly)

  24. Migrating Apps to Clouds • Candidate apps for migration have following C/C’s • Demand for resources vary with time • provisioning private data centers for peak wastes resources • Demand is not known in advance • Cannot provision private data centers; either too much waste (overprovisioning) or lost opportunities (underprovisioning) • Can leverage “cost associativity” • Using one machine for 100 hrs costs same as using 100 for 1 hr • Cloud migration transfers risk of miscalculating demandfrom user to cloud provider • Risk is mitigated by statistical multiplexing across multiple users

  25. Cloud Economics • Resources wasted in overprovisioning (left) and • Requests/services are rejected in under provisioning

  26. Cloud Economics • Cost-benefit analysis should consider • Variability of the demand • Cost of transferring data in/out of cloud • Utilization of private resources; typical server utilization 5-20% • Cannot have ~100% utilization as delay explodes • Cost of hardware drops during depreciation period (~3 years) • Cloud providers can reduce cost for customers • Human cost to manage private resources • Time to provision resources • Few minutes on clouds vs. weeks for private resources • Risk of early disposal of hardware • Termination of project, market change for product, … • extra cost

  27. Cloud Computing Model Cloud Applications Large-scale applications Cloud Services Domain-specific services Programming Models & Resource Management Virtualization, allocation, programming System Design Data center, storage system

  28. Data Centers

  29. Data Centers—Clusters

  30. Data Centers – Storage Hierarchy • Notice the differences in latency and bandwidth

  31. Data Centers—Software Infrastructure • Quite complex system to program • Many components • Different bandwidth and latency • Many failures •  Several tools and models to help • MapReduce • BigTable • Google File System • DryadLINQ • …

  32. Data Centers – Power Distribution

  33. Data Centers – Cooling

  34. PUE: Power Usage Effectiveness • PUE = Total building power / power in IT equipment • reflects quality of the datacenter building • Ideal to be 1.0 • Old data centers had PUE from 2.0 to 3.0 • Newer ones have PUE < 2.0 • Google reported PUE <1.10 in some recent data centers

  35. Power Overhead in Data Centers • Rough division of power overheads in data centers

  36. Data Centers – IT Power Consumption • No single component dominates power consumption

  37. Data Centers—Tiers • Tier I: • single path for power /cooling distribution, no redundant components • Tier II • adds redundant components (N + 1), improving availability. • Tier III: • Multiple power /cooling distribution paths but one active path • Provide redundancy even during maintenance, usually N + 2 • Tier IV: • two active power/cooling distribution paths, redundant components • Most commercial DCs are III and IV • Availability for II, III, IV: 99.75, 99.98%, 99.995%

  38. Summary • Demand of multimedia content is growing • QoS layers for end-to-end quality • Cloud computing … make computing utility • Candidate cloud apps have variable/unknown demand • Migrating to cloud, if feasible, may • reduce cost, • accelerate development/deployment, and • mitigate risk of estimating success/failure of new service/product

  39. References • Armbrust et al., Above the Clouds: A Berkeley View of Cloud. Computing, UCB/EECS-2009-28, Tech Report, February 2009 • Barroso and Holzle, The Datacenter as a Computer An Introduction to the Design of Warehouse-Scale Machines, 2009.