Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
COT 5611 Operating SystemsDesign Principles Spring 2010
Dan C. MarinescuOffice: HEC 439 BOffice hours: M-Wd 1:00-2:00 PM
Innovations are now full-circle
2
Factors affecting complexity
3
Complexity of computing and communication systems
New components
New applications
Interconnectivity + mobility, embedded
devices
Physical constraints
Larger segment of population using
the systems
Optimization of resource
consumption
TimingConstraints
Software follows hardware
0
10
20
30
40
50
60
Wind
ows 3
.1 (19
92)
Wind
ows N
T (199
2)Sola
ris (1
998)
Wind
ows 9
5W
indow
s 98
Wind
ows N
T 5.0 (
1998
)
RedHat
Linux
6.2 (
2000
)
RedHat
Linux
7.1 (
2001
)W
indow
s XP
Vista
Mill
ions
of
lines
of
sour
ce c
ode
Cheap → Pervasive
Pervasive → qualitative change
year
log
(peo
ple
per c
ompu
ter)
Slide from David Culler, UC Berkeley
Number crunching
EmbeddedSense/control
Word processingCommunication
Latency improves slowly
1
10
100
1000
1 2 3 4 5 6 7 8 9 10 11Year #
Impr
ovem
ent
wrt
yea
r #
1
Moore’s law (~70% per year)
DRAM access latency (~7% per year)
Speed of light(0% per year)
Heat is a problem
Recent Intel CPU Clock Rate
486
Pentium
PentiumPro
Pentium III
Pentium 4
Pentium 4 HT
mH
z
Lecture 210
Systems and the environment
System a set of interconnected components that has a an expected behavior observed at the interface with its environmentThe environment a critical component to be considered in the design of any systemThe systems we are concerned consist of
hardware andsoftware
Lecture 211
Two sources of complexity1. Cascading and interacting requirements
1.1 When the number of requirements grows then the number of exception grows. 1.2 The principle of escalating complexity:
Lecture 212
Two sources of complexity (cont’d)
1.3 Meeting many requirements with a single design the need for generality. Advice: avoid excessive generality.1.4 Requirement changes:
Example: the electric car produced by Tesla.
Lecture 113
Two sources of complexity (cont’d)
2. High performance2.1 Every system must satisfy performance standards. 2.2 The law of diminishing return the more one improves one performance metrics the more effort the next improvement will require
Lecture 214
Modularity for Coping with Complexity
Why does modularity reduce complexity we can focus on the interaction within one module/component.Example: assume that:
B - the # of bugs in a program is proportional with N , the number of statementsT- the time to debug a program is proportional with N x B thus it is proportional with N2
Now we divide the program in K modules each with N/K statements each:
The time to debug a module is proportional with (N/K)2
The time to debug the K modules is K x (N/K)2 = N2/KWe have reduced the time by a factor of K. Is that so?
Lecture 215
Abstractions and Coping with Complexity
Abstraction separation of the interface from the internals or specification from implementation
Example: you do not need to know how the engine of your car works in order to drive the car
Why abstractions reduce complexity because they minimize the interconnections between components.Observations:
Best division usually follows natural boundaries.The goal is defeated by unintentional or accidental interconnections among components.
Lecture 216
More about abstractions
• Abstractions are critical for understanding critical phenomena. Think about the abstract model of computation provided by the Turing Machine.
• Do not be carried away by abstractions. For example, often software designers think about an abstract computer and are not concerned about the physical resources available to them. E.g., the small display of a wireless phone.
Lecture 217
More about abstractions (cont’d)
The robustness principle be tolerant of inputs and strict on outputs.
Lecture 218
More about abstractions (cont’d)
The safety margin principle Keep track of the safety margin of the cliff or you may fall over edge!!
Lecture 219
Layering for Coping with Complexity
Layering building a set of successive functional entities with restricted communication patterns, a layer may only communicate with the layer below it and with the one above it.Examples: networking
Lecture 220
Hierarchy for Coping with Complexity
Hierarchical structures construct a large system from a small collection of relatively large subsystemsExamples:
CorporationsAn armyA computer is a collection of subsystems
Slides by Kaashoek & Morris
What went right?
Unbounded composibilityGeneral-purpose computers
Only need to make one thing fast
Separate architecture from implementationS/W can exploit new H/W
Cumulative R&D investment over years
Lecture 222
Names
Modularity along with abstraction, layering, and hierarchy allow a designer to cope with complexity;Names and addresses provide the means to connect modules. A system a bunch of resources, glued together with namesNaming allows the designer to:
Delay the implementation of some modules; use dummy onesReplace an implementation with another one.
Binding choosing an implementation for a moduleDelayed binding; use a place holder.
Lecture 223
Names and fundamental abstractions
The fundamental abstractions1. Storage mem, disk, data struct, File Systems, disk arrays2. Interpreters cpu, programming language e.g. java VM3. Communication wire, Ethernet
rely on names.Naming:
FlatHierarchical
Lecture 224
Computers a distinct species of complex systems
The complexity of computer systems not limited by the laws of physics distant bounds on composition
Digital systems are noise-free.The hardware is controlled by software
The rate of change unprecedentedThe cost of digital hardware has dropped in average 30% per year for the past 35 years
Lecture 225
Analog, digital, and hybrid systemsAnalog systems:
the noise from individual components accumulate and the number of components is limited
Digital systems:are noise-freethe number of components is not limitedregeneration restoration of digital signal levelsstatic discipline the range of the analog values a device accepts for each input digital value should be wider than the range of analog output valuesdigital components could fail but big mistakes are easier to detect than small ones!!
Hybrid systems e.g., quantum computers and quantum communication systems
Lecture 226
Computers are controlled by software
Composition of hardware limited by laws of physics.Composition of software is not physically constrained;
software packages of 107 lines of code exist
Abstractions hide the implementation beneath module interfaces and allow the
creation of complex softwaremodification of the modules
Abstractions can be leaky. Example, representation of integers, floating point numbers.
Lecture 227
Exponential growth of computersUnprecedented:
when a system is ready to be released it may already be obsolete. when one of the parameters of a system changes by a factor of
2 other components must be drastically altered due to the incommensurate scaling. 10 the systems must be redesigned; E.g.; balance CPU, memory, and I/O bandwidth;
does not give pause to developers to learn lessons from existing systemsfind and correct all errors
negatively affects “human engineering” ability to build reliable and user-friendly systemsthe legal and social frameworks are not ready
Lecture 228
Coping with complexity of computer systems
Modularity, abstraction, layering, and hierarchy are necessary but not sufficient.An additional technique iterationIteration
Design increasingly more complex functionality in the systemTest the system at each stage of the iteration to convince yourself that the design is soundEasier to make changes during the design process
Lecture 229
Iteration – design principles
Make it easy to changethe simplest version must accommodate all changes required by successive versionsdo not deviate from the original design rationalethink carefully about modularity it is very hard to change it.
Take small steps; rebuild the system every day, to discover design flaws and errors. Ask others to test it.Don’t rush to implementation. Think hard before starting to program.Use feedback judiciously
use alpha and beta versionsdo not be overconfident from an early success
Study failures understand that complex systems fail for complex reasons.
Lecture 230
Curbing complexity
In absence of physical laws curb the complexity by good judgment. Easier said than done because:
tempted to add new features than in the previous generationcompetitors have already incorporated the new features the features seem easy to implementthe technology has improvedhuman behavior: arrogance, pride, overconfidence…
Lecture 231
Resource sharing and complexity
A main function of the OS is resource sharing.Sharing computer resources went through several stages with different levels of complexity:
Many-to-oneOne-to-oneMany-to-manyPeer-to-peer
Lecture 132
33
HS – Homo SapiensPC- Personal Computer I
Host Guest
Host Guest
HostGuest
HostGuest
3434
Names and the three basic abstractions
Memory stores named objectswrite(name, value) value READ(name)file system: /dcm/classes/Fall09/Lectures/Lecture5.ppt
Interpreters manipulates named objectsmachine instructions ADD R1,R2modules Variables call sort(table)
Communication Links connect named objectsHTTP protocol used by the Web and file systemsHost: boticelli.cs.ucf.eduput /dcm/classes/Fall09/Lectures/Lecture5.pptget /dcm/classes/Fall09/Lectures/Lecture5.ppt
3535
MemoryHardware memory:
DevicesRAM (Random Access Memory) chipFlash memory non-volatile memory that can be erased and reprogrammedMagnetic tapeMagnetic DiskCD and DVD
SystemsRAIDFile systemsDBMS (Data Base management Systems)
3636
Attributes of the storage medium/system
Durability the time it remembers Stability whether or not the data is changed during the storagePersistence property of data storage system, it keeps trying to preserve the data
3737
Critical properties of a storage medium/system
Read/Write Coherence the result of a READ of a memory cell should be the same as the most recent WRITE to that cell.Before-or-after atomicity the result of every READ or WRITE is as if that READ or WRITE occurred either completely before or completely after any other READ or WRITE
3838
time
WRITE item A in cell M
AM
READ from cell M
AM
A A
Current READ/WRITE
Previous READ/WRITE
Next READ/WRITE
3939
Why it is hard to guarantee the critical properties? Concurrency multiple threads could READ/WRITE to the same cellRemote storage The delay to reach the physical storage may not guarantee FIFO operationOptimizations data may be buffered to increase I/O efficiencyCell size may be different than data size data may be written to multiple cells.Replicated storage difficult to maintain consistency.
4040
Access type; access time
Sequential access TapesCD/DVD
Random access devicesDisk
SeekSearch timeRead/Write time
RAM
41
Physical memory organizationRAM two dimensional array. To select a flip-flop provide the x and y coordinates.Tapes blocks of a given length and gaps (special combination of bits.Disk:
Multiple plattersCylinders correspond to a particular position of the moving armTrack circular pattern of bits on a given platter and cylinderRecord multiple records on a track
41
42
Names and physical addressesLocation addressed memory the hardware maps the physical coordinates to consecutive integers, addressesAssociative memory unrestricted mapping; the hardware does not impose any constraints in mapping the physical coordinates
42
Figure 2.2 from textrbook