260 likes | 366 Vues
CS 1651 Advanced Systems Software. Jack Lange Assistant Professor University of Pittsburgh. Course Objectives. Understand basic OS principals Understand historical background of OS design Gain practical experience with current systems x 86 and Linux
E N D
CS 1651Advanced Systems Software Jack Lange Assistant Professor University of Pittsburgh
Course Objectives • Understand basic OS principals • Understand historical background of OS design • Gain practical experience with current systems • x86 and Linux • Learn how modern systems are designed and operated • Gain some exposure to systems research
Administrivia • Instructor: Jack Lange • Email: jacklange@cs.pitt.edu • Office:Sennott Square #5407 • Office Hours: Weds 2-4PM
Communication • Course homepage • http://www.cs.pitt.edu/~jacklange/teaching/cs1651-f13/ • Announcements, clarifications, corrections • Additional resources for projects • Google Group • http://groups.google.com/group/pitt-cs1651-f13 • pitt-cs1651-f13@googlegroups.com • Private discussion group • Open venue for class discussions and questions • Based on email (Pitt addresses) • Email me if you want to use a different one
Class Details • Inverse of other graduate courses • Lectures on Tuesdays • Provide background • Supplemental readings assigned (Check Schedule!) • Presentations on Thursdays • Student presents selected research paper • Other students read and provide a review of the paper before hand
Grading • Grading • Participation (20%) • Final (30%) • 4 Projects (50%) • Late policy • Submit by midnight of the due date • 10% penalty for every day late
Projects • Work Individually • C is required • Highly Recommended: OS or having some familiarity with Unix systems programming, preferably in C or C++ • All projects are in C • BUILDING software is 50% of the grade of this class 1-9
Textbooks • None are “required” but highly recommended • Linux Device Drivers • Understanding the Linux Kernel
Reading papers and evaluating systems • What are the most important ideas: perhaps a combination of their motivations, observations, interesting parts of the design, or clever parts of their implementation. • What are the largest flaws; maybe an experiment was poorly designed or the main idea had a narrow scope applicability. Being able to assess weaknesses as well as strengths is an important skill for this course and beyond. • What is the relevance of the ideas today, potential future research suggested by the article, etc.
What makes a good presentation • What is the problem? • Why should people care about the problem? • How do you intend to address it (high level)? • How did you address it (w/ lowlevel details of interesting components)? • Why is your approach superior to others? • What are the shortcomings of your approach? • Did your approach work?
What makes a bad presentation • Lack of context • Reading from slides • Lots of text or data • Only focusing on technical details • Ignoring high level points
Why learn about Operating Systems? • Tangible reasons • Build or modify a real operating system • Administer and use system well • Tune application performance • Intangibles • Intrinsic curiosity • Understand how much of you computer system works • Gain/apply knowledge in other areas of Computer Science • Computer architecture and devices • Synchronization in programming languages • Data structures and algorithms • Performance analysis • Challenge of designing large, complex systems
What is an operating System • Not easy to define: Users Applications Operating System Hardware • OS: • Everything that isn’t an application or hardware • OS: • Software that abstract hardware into a useful form for applications • Standard Library • Resource Coordinator
First Function: Standard Library • Advantages of standard library • Allow applications to reuse common facilities • Make different devices look the same • Provide higher level abstractions • Challenges • What are the right abstractions?
Second Function: Resource Coordinator • Resource: “Anything valuable” • (e.g. CPU, memory, disk, network) • Advantages of resource coordinator • Virtualize resource so multiple users/applications can share • Protect applications from one another • Provide efficient and fair access to resources • Challenges • What mechanisms? • What policies?
What Functionality in OS? • No single right answer • Desired functionality depends on outside factors Technology Changes Expectations Users & Applications Operating System Computer Architecture • OS must adapt • Change abstractions provided to users • Change algorithms to implement those abstractions • Change low-level implementation to deal with hardware • Current operating systems driven by its evolution • Two distinct cases in history • Case 1: Computers are expensive • Case 2: Computers are cheap
History of the OS • Commercial systems (1950s) and HPC systems (early 1990s) • Enormousand expensive • Goal: Get the system working • Single operator/programmer/user runs and debugs at a time • OS functionality • Standard library -> No coordination of resources • Monitor that is always resident; transfer control to programs OS User Job Memory • Problem: Inefficient use of hardware • Performance Metrics • Throughput and Utilization
Batch Processing • Batch: Group of jobs submitted to machine together • Operator collects jobs; orders efficiently; runs one at a time • Role of OS: Same as before • Advantages • Amortize setup costs over many jobs • Keeps machine busy during a single users idle time • Disadvantage • User must wait for results until batch collected and submitted • If bug, receive memory and register dump; submit job again • Improve system throughput and utilization, but lose interactivity • Still the prevalent supercomputing model
Multiprogrammed Batch Systems • Spooling provides pool of ready jobs • Keep multiple jobs resident in memory • OS chooses which job to run • When job waits for I/O, switch to another resident job OS User Job 1 Memory User Job 2 • New OS functionality • Job scheduling policies • Memory management and protection (virtual memory) • Advantage: Improves throughput and utilization • Disadvantage: Still not interactive User Job 2
History of the OS: Phase 2 • Introduction of inexpensive, fast devices • Keyboards and monitors text editors and interactive debuggers • New set of performance trade-offs • Goal: True interactivity (via improved response time) • Time-sharing: Switch between jobs to give appearance of dedicated machine • Advantage • Users easily submit jobs and get immediate feedback • New OS functionality • More complex job scheduling, memory management • Concurrency control and synchronization
Personal Computers • Computers became cheap (late 70s, early 80s) • Dedicated machine per user • “Improved” functionality from OS • Remove time-sharing of multiple jobs • No protection • No virtual memory • OS becomes subroutine again • Conclusion: OS functionality changes with hardware and users OS User Job Memory
State of current systems • Large • 100k’s to millions of lines of code • 100-1000 person-years of work • Complex • Performance is important • Conflicting needs of different users • Poorly understood • System outlives any of its builders • Cannot remove all bugs • Behavior is hard to predict, tuning is done by guessing • Current trends • Multiprocessors • Networked systems • Cloud based systems
Message Passing • Explicit communication operations • Specify sender and receiver • Serialize data into a message • Transmit message over communication channel • Communication Channels • Sockets • E.g. Ethernet • RDMA (Remote direct memory access) • E.g. Infiniband • Shared memory • Becoming more prevalent on many-core architectures