Transcript Computer Classes: Why they form, and what`s new

Computer Industry Laws, Forces,
and Heuristics… Or,
Why computers are like they are
and are likely to be.
Gordon Bell
Computing Laws
Outline

Inventions, forces & laws
–
–
–
–


The two great inventions: Computer & IC
The force, quest and drive of cyberization
Resulting computer classes and
their supporting industries
The market support that drives it all
Technology to define new classes
Some inevitable new computer classes
Computing Laws
The two great inventions

The computer (1946).
Computers supplement and substitute
for all other info processors, including
humans
–
–

Computers are built from other
computers in a recursive fashion
Processors, memories, switching, and
transduction are the primitives
The Transistor (1946) and subsequent
Integrated Circuit (1957).
–
Computers are composed of a set of welldefined hardware-software
levels Laws
Computing
Everything cyberizable will be
in Cyberspace and covered by
a hierarchy of computers!
Continent
World
Body
Region/
Cars…
phys. nets
Intranet
Home…
Campus buildings
Fractal Cyberspace: a network
of … networks of … platforms
Computing Laws
Cyberization: interface to all
bits and process information

Coupling to all information and
information processors
 Pure bits e.g. printed matter
 Bit tokens e.g. money
 State: places, things, and people
 State: physical networks
Computing Laws
“
“
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
Computing Laws
Moore’s First Law

Transistor density doubles
every 18 months
60% increase per year
–
–

Exponential growth:
–
–

Chip density transistors/die
Micro processor speeds
1GB
128MB
1 chip memory size
( 2 MB to 32 MB)
8MB
1MB
128KB
8KB
1970
1980
bits: 1K 4K 16K 64K
The past does not matter
10x here, 10x there … means REAL change
1990
256K 1M 4M 16M 64M 256M
PC costs decline faster than any other
platform
–
–
2000
Volume and learning curves
PCs are the building bricks of all future systems
Computing Laws
Computer components must
all evolve at the same rate






Amdahl’s law: one instruction per second
requires one byte of memory and one bit
per second of I/O
Processor speed has evolved at 60%
Storage evolves at 60%
Wide Area Network speed evolves at 60%
Local Area Network speed evolved 26-60%
Grove’s Law: Plain Old Telephone Service
(POTS) thwarts speed, evolving at 14%!
Computing Laws
Bell’s law of computer class
formation to cover Cyberspace




New computer platforms emerge
based on chip density evolution
Computer classes require new
platforms, networks, and cyberization
New apps and content develop
around each new class
Each class becomes a vertically
disintegrated industry based on
hardware and software standards
Computing Laws
Bell’s Evolution Of
Computer Classes
Log price
Technology enables two evolutionary paths:
1. constant performance, decreasing cost
2. constant price, increasing performance
Mainframes (central)
Mini
WSs
PCs (personals)
??
Time 1/1.26 = .8
1.26 = 2x/3 yrs -- 10x/decade;
Computing
1.6 = 4x/3 yrs --100x/decade; 1/1.6
= .62 Laws
Network
Interface
Platform
Platform, Interface, & Network
Computer Class Enablers
“The
Mini &
Computer” Timesharing
Mainframe
PC/WS
Web browser,
telecomputer,
tv computer
tube, core, SSI-MSI, disk, micro, floppy, PC, scalable
drum, tape, timeshare
disk, bit-map servers,
batch O/S
O/S
display, mouse,
dist’d O/S
direct >
batch
terminals via
commands
WIMP
Web, HTML
POTS
LAN
Internet
Computing Laws
Bell’s Nine Computer Price Tiers
1$:
embeddables e.g. greeting card
10$:
wrist watch & wallet computers
100$:
pocket/ palm computers
1,000$:
portable computers
•
10,000$:
personal
computers (desktop)
100,000$:
departmental computers (closet)
1,000,000$:
site computers (glass house)
10,000,000$: regional computers (glass castle)
100,000,000$: national centers
Super server: costs more than $100,000
“Mainframe”: costs more than $1 million
Laws
an array of processors, disks,Computing
tapes, comm
ports
Computer Industry 1982
Solutions
Applications
OS
IBM DEC HP NCR
Computers
Processors
Computing Laws
Computer Industry 1995

Consult
Andersen, EDS, KPMG, Lante, etc.

Apps
Comshare, D&B, PeopleSoft, SAP

Apps
Microsoft, Lotus, WordPerfect, etc.

Dbases
Informix, Ingres, Oracle, Sybase,etc.

OS

Network

Periph

IBM, Compaq, DEC, Apple, many others
Computers

Micros

Solutions EDS, FDC, BTG, API, DataFocus, HFSI
Microsoft, Apple, Sun, Novell
Novell, Microsoft, Banyan
HP, Canon, Lexmark, Seagate
Intel, AMD, Motorola, others
Computing Laws
Economics-based laws
determine the market





Demand: doubles as price declines by 20%
Learning curves: 10-15% cost decline with 2X units
Bill’s Law for the economics of PC software
Nathan’s Laws of Software -- the virtuous circle
Metcalfe’s Law of the “value of a network”
Computing Laws
Software Economics: Bill’s Law
Price =



Fixed_cost
Units
+
Marginal _cost
Bill Joy’s law (Sun):
don’t write software for <100,000 platforms
@$10 million engineering expense, $1,000 price
Bill Gate’s law:
don’t write software for <1,000,000 platforms
@$10M engineering expense, $100 price
Examples:
–UNIX
versus Windows NT: $3,500 versus $500
–Oracle versus SQL-Server: $100,000 versus $6,000
–No spreadsheet or presentation pack on UNIX/VMS/...

Computing Laws
Commoditization of base software and hardware
The Virtuous Economic Cycle
that drives the PC industry
Standards
Computing Laws
Nathan’s Laws of software
1. Software is a gas. It expands to fill the
container it is in
2. Software grows until it becomes limited by
Moore’s Law
3. Software growth makes Moore’s Law
possible
4. Software is only limited by human ambition
and expectation
…GB: and our ability to cyberize I.e. encode
Computing Laws
Metcalf’s Law
Network Utility = Users2

How many connections can it
make?
–
–
–
–

1 user: no utility
100,000 users: a few contacts
1 million users: many on Net
1 billion users: everyone on Net
That is why the Internet is so “hot”
–
Exponential benefit
Computing Laws
The Virtuous Cycle
that drives the BW quest
Internet
(IP)
ubiquity
Computing Laws
Future Telecom Industry
Applicatio
ns
Ericsson, Aspect, Nortel, Octel, others
Microsoft, Delrina, many others
Applicatio Informix, Microsoft, Oracle, Sybase, others
ns
Microsoft, Apple, Sun, Novell, LINUX
Databases Ericsson, Nortel, Bay, 3Com, Fore, others
OS
Switching
Computers
DSP
Compaq, DEC, Dell, IBM, many others
Dialogic, NMS, Rhetorex, others
Intel, AMD, Motorola, others
Computing Laws
Hardware technology:
processing, memory,
networking, and new
interfaces enable the new
computers
Computing Laws
1. We get more
Computing Laws
Some changes by 2001


256 Mbit (32 Mbyte chip with computer)
LSI Logic is “System on a chip” co.
–
–


Mbit bandwidth will be like ISDN today
New networks will form to ferry us
amongh the “Islands of Cyberspace”
–

64 M gates (>100 M transistors) today
Embeddable, low cost products (e.g.
cameras, instruments) with processing,
memory, net, I/O
PC, phone, fax (unfortunately), pager,
radio/cell phone, home stuff, info appliances
Computing Laws
Cerf: “IP on everything.”
Extrapolation from 1950s:
20-30% growth per year
Tera
Giga
Storage
Backbone
Processing
Memory
??
Mega
Kilo
1
1947
Telephone Service
17% / year
1957
1967
1977Computing
1987 1997Laws
2007
National Semiconductor Technology
Roadmap (size)
10000
0.35
Memory size (Mbytes/chip) & Mtransistors/ chip
Mem(MBytes)
0.3
Micros Mtr/chip
Line width
1000
0.25
0.2
100
0.15
0.1
10
0.05
1
0
1995
1998
2001
2004
2007
2010
Computing Laws
National Storage Technology
Roadmap (size, density, speed)
100000
100000
3 .5 " Ca p . ( B y te s )
1 .3 " Ca p . ( B y te s )
10000
B its /s q . in .
10000
Da ta - r a te ( B y te s /s )
1000
1000
100
100
10
10
1
1
1995
2000
Computing
Laws
2005
Communication rate(t) in log10(Kbps)
10
1 Gb
???
9
8
SAN/backpanels
7
LAN
1 Mb 6
???
WAN
5
POTS @
17%/year
4
1 Kb
3
2
ISDN
POTS
1965
1975
Computing
1985
1995
Laws
2005
Microprocessor performance
100 G
Peak
Advertised
Performance
(PAP)
Real Applied
Performance
(RAP)
41% Growth
10 G
Giga
100 M
10 M
Moore’s
Law
Mega
Kilo
1970
1980
1990Computing
2000
2010
Laws
Gains if 20, 40, & 60% / year
60%=
Exaops
1.E+21
1.E+18
40%=
Petaops
1.E+15
20%=
Teraops
1.E+12
1.E +9
1.E+6
1995
2005
2015
2025
2035
2045
Computing Laws
New overtakes old
Computing Laws
Processor performance… also
for mainframes and supers
1000
100
10
Bipolar
processors
9000
RISC shift
1
VAX
0.1
CMOS
microprocessor
0.01
1970
1975
1980
1985
1990
1995
2000
Computing
Laws
Things get cheaper
Computing Laws
Exponential change of 10X per
decade causes real turmoil!
100000
10000
8 MB
1 MB
Timeshared
systems
1000 256 KB
100
$K
10
64 KB
16 KB
1
0.1
0.01
1960
Single-user
systems
1970
1980
1990
2000
Computing Laws
VAX Planning Model 1975:
I didn’t believe it

The model was very good
–

Costs declined > 20%
–


1978 timeshared $250K VAXen
cost about $8K in 1997!
users get more memory than predicted
Single user systems didn’t come
down as fast, unless you consider
PDAs
VAX ran out of address
bits!
Computing
Laws
Newer & cheaper always wins?
… if it weren’t for the Law of Intertia
Old
Old
New
New
Computing Laws
“The mainframe is dead!
… and for sure this time!”
P
R
I
C
E
Mainframe
Server
PC
Computing Laws
The law of data and program
inertia sustains platforms!




The investment in programs and processes to
use them, and data exceed hardware costs
The cost to switch among platforms e.g. IBM
mainframe, VMS, a VendorIX, or Windows/NT is
determined by the data and programs
The goal of hardware suppliers is
uniqueness to differentiate and lock-in
The goals of software/database suppliers are:
to differentiate and lock-in and
operate on as many platforms as possible
in order to be not tied to a hardware vendor
Computing Laws
Will the need for high volume,
higher performance micros
aka PCs continue?




Speech... but some of that power will be
embedded in appliances
Video requires extra-ordinary power,
especially to “understand”
Video servers!
The explosion of stored everything e.g.
photos, voice, video, requires more
memory and processing
Computing Laws
It’s the near-term platforms, stupid!
(multimedia is finally happening)

Text & 2D graphics -->> images, voice, & video

The WEB: being anywhere and doing anything
Disk sizes and cost c1998
– $50-100 / GB
– 4 GB standard; CD-R; and 20-40 GB magneto-optic R/W
Document, picture, and video capture and compression
– 10,000 to 250,000 pages / GB; 10,000 pictures / GB
– 40-400 books / GB or $0.25-2.50 / book
– Plethora of Video & digital cameras everywhere!
Voice and video compression*
– 250 hours / GB voice
– Stamp size-VHS: 12-50 hours / GB; DVD / HDTV: 0.5 hr / GB
Audio: Surround sound that is part of V-places
Ubiquitous access: NetPC, WebTV, web & videophones





Computing Laws
*Because there’s limited bandwidth!
What if could or when can we
store everything we’ve:
read/written,
heard, and
seen?
Computing Laws
“
“
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
Computing Laws
Computing Laws
Computing Laws
All those photos
Computing Laws
10X in 40 years
(6% per year)
Computing Laws
Library Volume Growth
10X in 150 years
Computing Laws
Some bits at Library of
Congress
Scanned LC
1PB
assumes 6B pages
 13M photos
13TB
 4M maps
200TB
 500K movies
500TB
 3.5M recordings 2,000TB


5 Bpeople or 2 GB per person
Computing Laws
Other bits per year
Cinema
5K
200TB
 Images (all) 52G
520PB
 Broadcast
1500st 200/10PB
 Recordings 100K 60TB
 Telephone
500Gmin 400PB
 videotape???

Computing Laws
Estimate of 1998 storage ships
http://www.lesk.com
Disks
25B
 Raid
13B
 Optical
0.5B
 Jukebox
5B
 Tape
10B
 Tape stack 2B

250PB
65PB
25PB
250PB
10,000PB -10EB
2000PB - 2EB
Computing Laws
Computing Laws
Static information storage sizes
Documents
business card
page or fax
snapshot
350 page book
image
5K
100 K
3M
25 M
4 drawer file 20Kp 100M
compressed
500
4K
100 K
2M
#/GB
200K;2M
10K;250K
10,000
40;500
10M
10;100
Computing Laws
Storing all we’ve read, heard, & seen
Human data-types
read text, few pictures
/hr
200 K
/day (/4yr)
2 -10 M/G
/lifetime
60-300 G
speech text @120wpm
speech @1KBps
43 K
3.6 M
0.5 M/G
40 M/G
15 G
1.2 T
video-like 50Kb/s POTS
video 200Kb/s VHS-lite
22 M
90 M
.25 G/T
1 G/T
25 T
100 T
video 4.3Mb/s HDTV/DVD 1.8 G
20 G/T
1P
Computing Laws
Some future computers and
networks
Computing Laws
Some predictable computers,
networks, & industries









Something NON-predictable
System-on-a-chip industry, including WINS
(Wireless Integrated Network of Sensors)
Digital still and video cameras
Dis-integrated telephony (gateways, IP dialing)
The “nc” (NC for LANs, WebTV, WebPhone)
Videophones become ubiquitous
Scalable Network And Platforms
Telework & Home Area Nets: homes, SoHos
Body Area Nets: “on body”,
“GuardianLaws
Angel”
Computing
2001 and the web will be
about as it is today…NOT

Bet: At least some appliance will be
available and selling at the rate of 2M
units per year averaged over the last
quarter of 2000 will have been
introduced that no one has predicted
at no 1997 conference about the
future of the Internet, excluding
cameras, television, and telephones
that access the web.
Computing Laws
Larry Ellison: NCs will outsell
PCs 9:1 by 2000.


NCs include those embedded in TV sets,
phones, and used as PC alternatives.
Bet: While the combined set of
computers connected to the web (e.g.
instruments, cameras, tv sets,
appliances, printers, phones) may be
greater than pure PCs, the number of
person-driven access devices that are
NOT PCs will be less than 1:1 by the end
of 2000.
Computing Laws
SNAP: Scalable Networks and
Platforms
Standard (I.e. commodity) hardware
 SAN (System Area Network)
alternatives
 Common operating system for
platform, reducing vendor and
customer costs
 Cluster technology

Computing Laws
Scaling dimensions include:




reliability… including always up
number of nodes
– most cost-effective system built from
best nodes… PCs with NO backplane
– highest throughput distributes disks
to each node versus into a single
node
location within a region or continent
time-scale I.e. machineComputing
generations
Laws
SNAP Systems circa 2000
Portables
Mobile
Nets
Wide-area
global
ATM network
Person
servers
(PCs)
Telecomputers
aka Internet
Terminals
???
TC=TV+PC
home ...
(CATV or ATM
or satellite)
Local &
global data
comm
world
ATM & Ethernet:
PC, workstation,
& servers
Legacy
mainframe &
minicomputer
servers & terminals
scalable computers
built from PCs + CAN
A space, time (bandwidth),
generation, and reliability
scalable environment
Centralized
& departmental
servers built from
PCs
Computing Laws
Do any hardware systems
vendors with proprietary
microprocessors and O/Ss
see the change?
Probably not. The web
business is masking it!
Computing Laws
Telework = work + telepresence
“being there while being here”

The teleworkplace is just an office with limited
–
Communication, computer, and network support!
Team interactions for work! Until we understand in situ
collaboration, CSCW is a “rat hole”!
– Serendipitous social interaction in hallway, office coffee
place, meeting room, etc.
– Administrative support for helping, filing, sending, etc.
Telepresentations and communication
Computing environment … being always there,
administrivia,
phones,
information (especially paper) management
SOHOs & COMOHOs is a high growth market
–



Computing Laws
Teleworking CW 9/1/97



15% 2 yr increase, 11 Mpeople, avg. 19 Hr/wk
50% in U.S.; 22% have policies on screening,
worker expectations, IP etc. protection,
liability
Are telecommuters more productive?
–
–
–
–

30% yes
50% same
4% no
16% don’t know
Are telecommuters more accessible?
–
–
–
–
13% yes
40% same
40% more
7% don’t know
Computing Laws
Steve Mann
in
Cyberspace
Computing Laws
CMU wearable computers
Computing Laws
Medtronics
Implanted
Cardioplastic
Computing Laws
The growth of the computer industry
(Gordon’s swag 12/97)













Machine class 1992
1995
1998
2001
Handheld/mobile
>
>
>>
PC (portables)
>
>
>
>
PC (desktop)
=
>
=
=
Telecomputer
>
>>
Network Computer >
>>
TC (TV Computer) na
na
>>
>>
Workstation
=
=
<
<
VendorIX server
>
>>
=
<
Mainframe
<
<
<
<<
Super
=
<
<
<<
Scalable PCs
=
>
>>
>>
Laws
= 0-10%, >10-20%, >> 20-30%; <Computing
-10%