Six Laws of Computing

							Laws of Cyberspace
Jim Gray Microsoft Research with help from Gordon Bell, Nathan Myrvold and laws by Bell, Moore, Gates, Joy, Gilder, Grove, Grosch, Metcalf, Mryvold,
Talk presented 10/9/98 at International University, Bruchsal Germany http://research.Microsoft.com/~Gray/talks/Laws_of_Cyberspace.ppt
1

Computer Industry Laws (rules of thumb)
• • • • • • • • • • • • • Metcalf’s law Moore’s First Law Bell’s Computer Classes (7 price tiers) Bell’s Platform Evolution Bell’s Platform Economics Bill’s Law Software Economics Nathan’s 4 Laws of Software Gilder’s Law of the Telcosom. Grove’s law (1 and 2) Moore’s second law Is Info-Demand Infinite? The Death of Grosch’s Law

2

1. We get more

3

2. New overtakes old

4

3. Things get cheaper

5

4. Newer & cheaper wins?

Old Old New New

6

Metcalf’s Law Network Utility = Users2
• How many connections can it make?
• 1 user: no utility • 1K users: a few contacts • 1M users: many on net • 1B users: everyone on net

• That is why the Internet is so “hot”
• Exponential benefit

7

Moore’s First Law
•XXX doubles every 18 months 60% increase per year 1GB –Micro Processor speeds 128MB –chip density 1 chip memory size 8MB ( 2 MB to 32 MB) –Magnetic disk density 1MB –Communications bandwidth 128KB
WAN bandwidth approaching LANs
1980 1990 2000 •Exponential Growth: bits: 1K 4K 16K 64K256K 1M 4M 16M64M256M –The past does not matter –10x here, 10x there, soon you're talking REAL change. •PC costs decline faster than any other platform –Volume & learning curves –PCs will be the building bricks of all future systems 8KB 1970

8

Bumps in the Moore’s Law Road
1000000
$/MB of DRAM

• DRAM: • 1988: US Anti-Dumping rules • 1993-1995: ?? price flat

10000 100 1 1970

1980

1990

2000

• Magnetic Disk • 1965-1989: • 1989-1996:

$/MB of DISK

10,000

10x/decade 4x/3year! 100X/decade

100

1
.01 1970

1980

1990

2000
9

National Semiconductor Technology Roadmap (size)
10000
Mem(MBytes)

0.4 0.35 0.3 0.25

Memory size (Mbits/chip) & Mtransistors/ chip

Micros Mtr/chip

1000

Line width

100

0.2 0.15

10

0.1 0.05

1 1995 1998 2001 2004 2007 2010

0

10

National Storage Technology (disks) Roadmap (size, density, speed)
100,000 3.5" Cap. (MBytes) 1.3" Cap. (MBytes)

10,000

1,000

100

10

1 1995 2000 2005
11

Gordon Bell’s 1975 VAX planning model... He didn’t believe it! System Price = 5 x 3 x .04 x memory size/ 1.26
5x: Memory is 20% of cost 3x:DEC markup .04x: $ per byte He didn’t believe: The projection 500$ machine He couldn’t comprehend implications
100,000.K$ 10,000.K$ 1,000.K$ 100.K$ 10.K$ 1.K$ 0.1K$ 0.01K$ 1960 16 KB 1970 64 KB 1980 256 KB 1990 1 MB 2000 8 MB

(t-1972)

K$

Costs declined > 20% Single user systems didn’t come down as fast, unless you consider PDAs VAX ran out of address bits!

12

Gordon Bell’s Seven Price Tiers
• • • • • • • 10$: 100$: 1,000$: 10,000$: 100,000$: 1,000,000$: 10,000,000$: wrist watch computers pocket/ palm computers portable computers personal computers (desktop) departmental computers (closet) site computers (glass house) regional computers (glass castle)

SuperServer: Costs more than 100,000 $ “Mainframe” Costs more than 1M$ Must be an array of processors, disks, tapes comm ports
13

Bell’s Evolution of Computer Classes
Technology enable two evolutionary paths: 1. constant performance, decreasing cost 2. constant price, increasing performance
Mainframes (central) Log Price

Minis (dep’t.)
WSs PCs (personals) Time 1.26 = 2x/3 yrs -- 10x/decade; 1/1.26 = .8 1.6 = 4x/3 yrs --100x/decade; 1/1.6 = .62

??
14

Everything cyberizable will be in Cyberspace and covered by a hierarchy of computers! Continent
Body Region/ Cars… phys. nets Intranet Home… Campus buildings

World

Fractal Cyberspace: a network of … networks of … platforms
15

Many little beat few big
$1 million
1 MM
3

$100 K

$10 K
Pico Processor

Micro

Mainframe

Mini

Nano 1 MB

10 pico-second ram

10 nano-second ram 100 MB 10 GB 10 microsecond ram 10 millisecond disc 100 TB 10 second tape archive 1 TB

14"    

9"

5.25"

3.5"

2.5" 1.8"

1 M SPEC marks, 1TFLOP

Smoking, hairy golf ball How to connect the many little parts? How to program the many little parts? Fault tolerance?

106 clocks to bulk ram
Event-horizon on chip VM reincarnated Multi-program cache, On-Chip SMP
16

Gordon Bell’s Platform Economics
• Traditional computers: Custom or Semi-Custom high-tech and high-touch • New computers: high-tech and no-touch
100000 10000 1000 100

units

$
Price (K$) Volume (K) App price

10
1 0.1 0.01 Mainframe WS Browser

Computer type

17

Software Economics: Bill’s Law

Fixed _ Cost Price   Marginal_Cost Units
• Bill Joy’s law (Sun):
• Bill Gate’s law:
Don’t write software for less than 100,000 platforms. @10M$ engineering expense, 1,000$ price

Don’t write software for less than 1,000,000 platforms. @10M$ engineering expense, 100$ price • Examples: • UNIX vs NT: 3,500$ vs 500$ • Oracle vs SQL-Server: 100,000$ vs 6,000$ • No Spreadsheet or Presentation pack on UNIX/VMS/... • Commoditization of base Software & Hardware
18

Software Economics
• An engineer costs about Microsoft: 9 B$ Profit R&D 150 k$/year 24% 16% • R&D gets [5%…15%] of budget Tax SG&A • Need [3M$…1M$] revenue 13% 34% per engineer Product&Service
13%

Intel 16 B$
Profit 22%

IBM: 72 B$
SG&A 11% Profit Tax 6% 5%

Oracle: 3 B$
Profit 15% Tax 7% Product& Services 26%

R&D 8%

R&D 8%
SG&A 22%

R&D 9%
SG&A 43%
19

Tax 12%

Product&Service 47%

Product&Service 59%

Grove's Law The New Computer Industry

• Horizontal integration is new structure • Each layer picks best from lower layer. • Desktop (C/S) market • 1991: 50% • 1995: 75%

Function Operation Integration Applications Middleware Baseware Systems

Silicon & Oxide

Example AT&T EDS SAP Oracle Microsoft Compaq Intel & Seagate

20

Bytes/$ DRAM
1000000 100000 10000

Bytes/$

1000 100 10 1

Doubling time 964 days Growth rate 30% per year
1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000

21

Nathan’s 1st Law of Software

Software is a gas!
It expands to fit the container it is in!

22

Windows NT Lines of Code
100,000,000

Doubling time 866 days

Growth rate 33.9% per year
10,000,000

1,000,000 7/92

2/93

8/93

3/94

10/94

4/95

11/95

6/96

12/96

6/97
23

Browser Code Growth (MB vs time)
100

Doubling time 216 days

Growth rate 221% per year
10

1

2/95

5/95

8/95

11/95

3/96

6/96

9/96

12/96

3/97

6/97

24

Nathan’s 2nd Law of Software

Software grows until it becomes limited by Moore’s Law
• Initial growth is rapid - like gas expanding (like browser) • Eventually, limited by hardware (like NT) • Bring any processor to its knees, just before the new model is out

25

Nathan’s 3rd Law of Software

Software growth makes Moore’s Law possible
• That’s why people buy new hardware economic motivator • That’s why chips get faster at same price, instead of cheaper • Will continue as long as there is opportunity for new software

26

Nathan’s 4th Law of Software

Software is only limited by human ambition & expectation
•It’s impossible to have enough •New algorithms •New applications and new users •New notions of what is cool

27

The Software Crisis!
• Von Neumann had trouble • Software is always in “crisis” • Is there some limit to complexity? • Will software ever grow up? • Will the crisis ever end?

Of course not!
28

The Perpetual Crisis •Panacea solutions •High level languages •Object oriented programming •Component software, ... •Benefits absorbed by rising expectations •Software will never be easy •Somebody will push the boundary
29

The Ultimate Computer
• Nathan’s Prognosis •Learning more about the brain every day •AI will happen •Computers with same power in 20 to 30 years • Brain has no Moore’s Law

30

Gilder’s Telecosom Law:
3x bandwidth/year for 25 more years
• Today: • 10 Gbps per channel • 4 channels per fiber: 40 Gbps • 32 fibers/bundle = 1.2 Tbps/bundle • In lab 3 Tbps/fiber (400 x WDM) • In theory 25 Tbps per fiber • 1 Tbps = USA 1996 WAN bisection bandwidth

1 fiber = 25 Tbps

31

God Loves Standards: That’s why he made so many of them.
1985
UNIX International

Microsoft DCOM based on OSF-DCE Technology DCOM and ActiveX extend it
Open software Foundation (OSF)

1990
Object Management Group (OMG)

Solaris

X/Open

1995
CORBA

Open Group

OSF DCE

NT
COM
32

Moore’s Second Law
•The Cost of Fab Lines Doubles Every Generation
(3 years) • Money Limit: hard to imagine 10 B$ line 20 B$ line 40 B$ line

$10,000

M$ / Fab Line

$1,000 $100 $10 $1 1960

• Physical limit: • Quantum Effects
at 0.25 micron now 0.05 micron seems hard 12 years, 3 generations

1970

1980

1990

2000

Year

• Lithograph: need Xray below 0.13 micron

33

Constant Dollars vs Constant Work •Constant Work:
• One SuperServer can do all the world’s computations.

• Constant Dollars:
• The world spends 10% on information processing • Computers are moving from 5% penetration to 50%
• 300 B$ to 3T$ • We have the patent on the byte and algorithm

34

Computer Industry Laws (rules of thumb)
• • • • • • • • • • • • • Metcalf’s law Moore’s First Law Bell’s Computer Classes (7 price tiers) Bell’s Platform Evolution Bell’s Platform Economics Bill’s Law Software Economics Nathan’s 4 Laws of Software Gilder’s Law of the Telcosom. Grove’s law (1 and 2) Moore’s second law Is Info-Demand Infinite? The Death of Grosch’s Law

35

“

There will always be plenty of things to compute ... With millions of people doing complicated things. memex … stores all his books, records, and communications, and ... can be consulted with speed and flexibility

“

”

” “ Matchbook sized, $.05 encyclopedia ” “ Speech to text ” Head mounted camera, dry photography “ ”
Vannevar Bush c1945
36

Kinds Of Information Processing
Point-to-Point
Immediate Time Shifted
conversation money

Broadcast
lecture concert

Net work Data Base

mail

book newspaper

Its ALL going electronic Immediate is being stored for analysis (so ALL database) Analysis & Automatic Processing are being added
37

Why Put Everything in Cyberspace?
Immediate OR Time Delayed Point-to-Point OR Broadcast
Network Locate Process Analyze Summarize Data Base
38

Low rent min $/byte Shrinks time now or later Shrinks space here or there Automate processing knowbots

Information At Your Fingertips™ Information Network™ Knowledge Navigator™

Databases:

• All information will be in an online database (somewhere) • You might record everything you
• read: 10MB/day, 400 GB/lifetime (8 tapes today) • hear: 400MB/day, 16 TB/lifetime (3 tapes/year today) • see: 1MB/s, 40GB/day, 1.6 PB/lifetime (maybe someday)

• • • • •

Data storage, organization, and analysis is a challenge. That is what databases are about DBs do a good job on “records” Now working on text, spatial, image, and sound. This needs lots of PROCESSING too.

39

Database Store ALL Data Types
• The Old World: – Millions of objects – 100-byte objects
People
Name Address

David Mike Won

NY Berk Austin People

• The New World: • Billions of objects • Big objects (1MB) • Objects have behavior (methods)
Paperless office Library of congress online All information online entertainment publishing business WWW & Internet Information Network, Knowledge Navigator, Information at your fingertips
40

Name Address Papers Picture Voice

David NY Mike Berk

Won Austin

Magnetic Storage Cheaper than Paper
• File Cabinet:
cabinet (4 drawer) paper (24,000 sheets) space (2x3 @ 10$/ft2) total 250$ 250$ 180$ 700$

3 ¢/sheet

• Disk:
• Image:

disk (4 GB =) 800$ ASCII: 2 m pages 0.04 ¢/sheet (80x cheaper)
200 k pages

0.4 ¢/sheet (8x cheaper) • Store everything on disk

41

Crossing the Chasm
New Market

product finds customers hard

No Product No Customers

Old Market

Boring Competitve Slow Growth

hard

Customers find product

Old Technology

New Technology
42

Billions of Clients
• Every device will be “intelligent” • Doors, rooms, cars, ... • Computing will be ubiquitous

43

Billions of Clients Need Millions of Servers
All clients are networked to servers
may be nomadic or on-demand
mobile clients

Clients
fixed clients

Fast clients want faster servers Servers
server

Servers provide data, control, coordination communication

Super Servers Large Databases High Traffic shared data
44

super server

The Parallel Law of Computing

Grosch's Law:

2x $ is 4x performance 2x $ is 2x performance Parallel Law:

1 MIPS 1$

1,000 MIPS 32 $ .03$/MIPS

Needs Linear Speedup and Linear Scaleup Not always possible

1,000 MIPS 1,000 $ 1 MIPS
1$
45

“The mainframe is dead! … and for sure this time!”
P R I C E

Mainframe Server PC

46

Useful Aphorisms
• There are no silver bullets.
Fred Brooks

• There is no such thing as a heterogeneous system.
Butler Lampson

• You know you have a distributed system when a computer you have never heard of prevents yours from working.
Leslie Lamport

• Hubris: the Greek word for “second system.”
Bob Stewart

• Software is like entropy, it weighs nothing, it is hard to understand, and it always increases.
Norman Augustine
47

Scaleable Systems BOTH SMP and Cluster

Grow Up with SMP 4xP6 is now standard

SMP Super Server

Grow Out with Cluster Cluster has inexpensive parts

Departmental Server Personal System

Cluster of PCs

48

SMPs Have Advantages
• Single system image easier to manage easier to program threads in shared memory, disk, net • 4x SMP is commodity • Software capable of 16x • Problems: • > 4 not commodity • scale-down problem (starter systems expensive)

SMP Super Server

Departmental Server Personal System
49

Clusters Have Advantages
• Clients and Servers made from the same stuff. • Inexpensive: • Built with commodity components • Fault tolerance: • Spare modules mask failures

• Modular growth • grow by adding small modules

50

Future SuperServer

4T Machine
Array of 1,000 4B machines 1 bips processors, 1 BB DRAM 10 BB disks, 1 tapes 1 Bbps comm lines A few MegaBucks

100 Tape Transports = 1,000 tapes = 1 PetaByte

1,000 discs = 10 Terrorbytes 100 Nodes 1 Tips

High Speed Network ( 10 Gb/s)

Challenge:
Manageability Programmability Security Availability Scaleability Affordability As easy as a single system

51