OS Background Knowledge

All backgroud information needed for learning OS. Will be updated countineously.

OS as a Resource Manager

• File-System Management
  • Create, deleting files and directories
  • Supporting primitives for manipulating files and directories
  • Mapping files onto mass storage
  • Backing up files on stable(nonvalatile storage media)
• Mass-Storage Management
  • Mounting and unmounting storage devices
  • Free-space management
  • Storage allocation
  • Disk scheduling
  • Partitioning
  • Protection
• I/O System Management - via I/O subsystem
  • A memory-management component with - buffering, caching, and spooling
  • Make general uniform device-friver interface
• Cache management
  • info is kept in some storage system. Check cache before need a particular piece of info
  • Cache manage in a multiprogram system becomes tricky and even more in a multiprocessor environment
  • Cache size and cache replacement policy

Evolution of OS

1. Serial Processing
2. Simple Batch Systems
3. Multi-programmed Batch Sysems
4. Time-Sharing Systems

Difference

	Batch Multi-prgramming	Time Sharing
Principle Objective	Max processor use	Min response time
Source of intructions to operate system	Job control language JCL instructions provided with the job	Commands entered at the terminal

Dual Mode Operation

• Kernal Model : Supervisor mode, system mode, privieged mode
• User Mode

A mode bit kernel 0 or user 1 is used to indicate the current mode.

Kernel Data Structure

1. Array
   • Each lement can be accessed directly
2. List: Sequence of data value
   1. Singly linked list
   2. Doubly linked list
   3. Circular linked list: last point to the first
3. Stacks
4. Queues
5. Tree / Binary Trees
6. Hash functions
7. Bitmaps: available table for resources

Linux Kernal Data Structure

• Linked-list data structure details are in the include file <linux/list.h>
• A queue in linus is know as kfifo, with implementation in kfifo.c file of kernel directory
• Implementation of the balanced binary search tree implementation using red-black trees can be found in the include file <linux/rbtree.h>

Virtualization

• Emulation - simulates computer hardware in software
• Virtualization - an OS that is natively compiled for a particular CPU architecture runs within another OS also native to that CPU

Distributed System

Network

Communication protocal

TCP/IP is the most common

Distance

• Local Area Network (LAN): room
• Wide Area Network (WAN): building
• Metropolitan-Area Network (MAN): city
• Personal-Area Network (PAN): BlueToothand 802.11 or other wireless

Communication Media

• Copper wires
• Fiber links
• Wireless trans between satellites
• Microware dishes
• Radio
• Infra-red
• ….

Loader and linker

• Relocatable object file - source code in complied into object files by the compiler and is designed to be loaded into any physical memory localtion
• Linker combines the object files into a single binary executable file
  • also links the library object modules nedded by the program
• Loader brings the executable file into memory to run the program
  • Relocation assign final addresses to program parts and adjusts code and data in program to match those address
• Modern negeral-purpose systems dont link library into executables
  • but, hynamically linked libraries are loaded as needed, shared by all that use the same version of that same library

Computer System Architechture

• Most use single general-purpose processor
  • some use special-purpose
• Multiprocessor System
  • Parallel or tightly coupled systems
  • offer
    • increased throughput
    • economy of scale
    • increased reliability
  • Two Type
    • Asymmetric Multiprocessing (ASMP) - each processor is assigned a special task
    • Symmetric Multiprocessing (SMP) - each processor performs all tasks

Asymmetric	Symmetric
All processor not equally	All equal
One master processor carries out the taskes of OS	Carried out by all processor
Processor cannot direct communicate with other processor (only via master)	All can communicate by a shared memory
Processes are categorized as master-slave	A ready queue is used for taking up processes
Cheaper	Relatively costly
Easier to design	Complex

Dual Core Design

• A processor have two cores
• Each core has its own level 1 cache, but shared level 2 cache

Clustered Systems

• like multiprocessor systems, but multiples systems are working together in this case
• combine cheap multiple processing unit to provide high performance and efficient systems
• High-acailability service - continue even if one or more systems in the cluster fail
• Each node can monitor one or more of other nodes, in case of failure
• Users and clients see only a brief interruption of service
• Stotage shared by storage-area network SAN
• Types:
  • Asymmetric clustering - one machine in hot-standby, other run application. If server fails, hot-standby becomes active server
  • Symmetric clustering - multi nodes run app and monitor each other
• Use:
  • High-performance computing HPC
  • Fault tolerant mass data storage
  • Graceful degradation
• OS design considerations and issues
  • Load balancing
  • Fair share of resources
  • Communication overhead

Hadoop and Big Data

• High availability distributed object oriented platform - an open source sfwr framework for storing data and running app on clusters of commodity hardware
• In Java and run on Linux
• Provide massive storage for data, enormous processing power an handle huge # of concurrent tasks
• Not an OS, but frameware developed on top of Linux, giving it a distribured OS capability
• Its app can be written in many language
• Use it for big data

Computer Environment

1. Traditional Computing
2. Mobile Computing
3. Client - Server Computing
   1. Compute-server systems
   2. File-server system (most web server)
4. Peer to Peer P2P Computing
   • Every node is a server and client
5. Cloud Computing
   • Public Cloud
   • Private
   • Hybrid
   • SaaS
   • PaaS
   • IaaS
6. Real-time Embedded Systems

Operating System Structure

1. Monolithic Strcture - Tightly coupled system - shared memory
   • All function of nernal placed in single, static binary file that run in single address space
   • Run as a single program
   • Original UNIX
   • Typical use glibc standard C lib to communicate with kernel
   • Linux has similar kernel but has a modular design that allows kernel to be modified during run time
2. Layered Approach - lossly couples system - distributed memory
   • brouen into number of layers
   • Layer 0 (botton layer) is hardware, Highest layer N is the user interface
   • Mth layer consists of data structures and function that can be invoked by higher layers
   • Advantage
     • Simplicity
     • Debugging
   • Each layer Hides existance of data struc, operations, and HW from higher level
   • Can be used in computer network (like TCP/IP) and web app
3. Microkernels
   • 
   • Noneseential component are removed from the kernel
   • Advantage
     • easier extension
     • easily ported from one hardware to other
     • more security and reliability
   • Drawbacks
     • Copying of messages and switching between processir
4. Modules
   • UseLoadable Kernel Modules LKMs - best current method for OS design
   • Kernel provide core services
   • Futher service and be linked dynmically while kernel running
   • Linking services does not require the recompilation of kernel
5. Hybrid System
   • Different structure combine