Transcript Chapter 9 - 1981 to 1995

Chapter 9 - 1981 to 1995
Workstations, UNIX
& the Net
1
Next Step - Workstations

Inexpensive microprocessor



Cost less than mini; more than PC
Main Features



Motorola 68000
UNIX
Extensive Networking Capabilities
Idea: Attach these to mainframe rather
than dumb terminal
2
Apollo - First Workstation






Bill Poduska, from Prime Computer
Domain: own OS and NW system
$40,000
Used for CAD & engineering
Mid-1980 - sold 1,000
1989- bought by H.P.
3
Sun Microsystems


1982- founded by
Vinod Khosla
Also Bill Joy


Stanford University
Network Workstation


Grant - UNIX
Andy Bechtolsheim
June 1982- SUN-2,
$20,000

Berkeley UNIX
First SUN Workstation - 1983
4
UNIX


AT&T Bell Labs, NJ; Ken Thompson,
Dennis Richie
Not a complete OS


Due to legal actions




Set of tools to manipulate & share files
AT&T couldn’t sell for profit
Universities got license for cheap
Commercial could also buy
Open Source
5
The UNIX Journey

Developed in New Jersey


To easily share files; Very frugal
Not for masses;

Univ. of Illinois-Champagne-Urbana

U.C. Berkeley


Extensively rewritten
Bill Joy

Took it to SUN
6
UNIX and Universities




Cheap source code
Written in C; run any machine with C
compiler
Free to modify code - and they did
Berkeley Software Distribution (BSD)
UNIX

1978-Joy offering tapes cheap
7
Universities (cont.)
1980 - ARPA backed BSD
 Version 4.2




Network Protocol TCP/IP
ARPA promoted TCP/IP
Forever linked UNIX & Internet
8
UNIX * Miscellaneous

VAX - Berkley UNIX w/ TCP/IP



Helped transform ARPANET to Internet
Vulnerable to viruses
Never really challenged Windows

Not even LINUX, yet
9
Vax Strategy - 1980’s




Offer single architecture (VAX) with single
OS (VMS) in solitary or networked
configurations ranging from desktop to
mainframe capability
Networking – Ethernet - from Intel & Xerox
“The network is the computer.”
Several Modes: 11/780, 11/750, MicroVAX
II, 8600 (Venus), 9000
10
Vax Strategy Risks



Similar to IBM’s “betting the
company”
Had to supply customers with
everything without seeming to
change too much
Entire line had to be high in quality
11
Risks (cont.)

Stop marketing own competing H.W.


Public outcry over PDP-10 &
DECtape


PDP-10- Outdated
Phase out an announcement
Historical Perspective- Pg. 186
12
Vax Strategy Results

Did not stick with it






1982 - 3 incompatible machines (not IBMPC
compatible - fatal)
Strategy went well through 1980’s
1987 stock market crash
Competition - UNIX workstations & IBM PC
DEC couldn’t recover #2 position
Final blow: Did not develop current
architecture
13
RISC


Reduced Instruction Set Computer
IBM-360, DEC VAX






Complex Instruction Set Computer (CISC)
200+ instructions, each
Due to slow access core memory
Due to immature compilers
Trying to close “English Instruction” gap
Cheap ROM allowed low cost of CISC
14
RISC- More #1

John Cocke, IBM “wild duck”


Experimental: IBM 801, 1979




Did not make market
1980 - Berkeley- RISC Project
1981- Stanford


Improved technology  believed smaller set of
instructions with more loads & stores would be faster
than 370
MIPS (Millions of instructions per second)
Skepticism outside university environment
Everything else booming - so why change?
15
RISC - More #2

1987- SUN SPARC- RISC Chip




Scalable Processor Architecture
Overcame Skepticism
RISC improved microprocessors speeds
faster than mainframe & miniprocessors were improving
Sun Licensed SPARC to others


Hoped it would become the standard
But would not be profitable
16
RISC – More #2 (cont.)

MIPS computer systems





Stanford MIPS project
DEC bought RISC chip for workstation
Silicon Graphics
1990- IBM R/6000
1990’s early: IBM & Apple

Power PC, Motorola Chip
17
Workstation vs. PC




RISC Architecture
Scientific & Engineering Apps.
Networking (Ethernet)
Cost
18
Ethernet



Developed @ Xerox PARC, 1973
Robert Metcalfe & David Boggs
Metcalfe



At MIT in 1969- helped connect PDP-10 to
ARPNET – to do same in ‘72 at PARC
Focus @ PARC was local networking
PARC Local Network


Data General minis in star technology
Expensive, inflexible, not robust
19
ALOHAnet




To connect among Hawaiian
Islands
Radio Signals  Wireless
Packets of 1000 bits; address of
recipient attached to head of each
message
Computers turned to UHF frequency
& listened for packets
20
Network Features #1


Radio (medium) was passive
Computers (Nodes) did the work


“Ether”- invisible medium


Process, queue, route
Replaced by coaxial cable
New Computer just taps into cable
21
Network Features #2


Computer “listens” before sending
Collision: random pause, try again



If many collisions, send less frequently
Math analysis showed would work
1974- Running @ 3 million bps

Arpanet 50 (telephone) - kilobits/sec
22
Ethernet Impacts




Speed changed relationship between
small and large computers
1st affected workstations, then PC market
DEC, INTEL, Xerox: accepted as
standard for VAX
DOS/ Early PC chips - not well suited
for networking
23
Apple  PC’s


With Lotus 1-2-3, Word Processing,
& dBase III, IBM compatibles began
to replace Apples & Word
Processors in office environment
Less expensive clones
24
“Personal” Computing in Business

Employees had personal SW



Became problem for I.S. people


Not in line with business goals
Some sw not very good
So LAN’s helped to “control” technology
Irony: networking made it not so
personal
25
Novell





Networking practical after 80386
1989 - had half business
Complex, expensive, overlaid DOS
 File server with software
Not as good a UNIX networking with
workstations
Backups, messaging, sharing
26
Internet


LAN’s provided access to Internet
Key features





Descendent of ARPANET
Packet switching
No dedicated line necessary
TCP/ IP- standard protocol
Open to public, commercial
27
Internet Success


ARPA’s support; adoption of
TCP/IP in 1980
TCP/IP inclusion into Berkeley
UNIX


Not proprietary
Rise in number of LAN’s
28
Success (cont.)


Ethernet Speeds
Grove’s Law



Telecommunication bandwidth doubles
every 100 years
Cable, etc. have improved
“Last Mile Problem”
29
Internet Before WWW

Arpanet- goal was resource sharing



Groups


FTP, Telnet: had to know location of information
Email - did emerge
Bulletin Boards, Discussion Groups, Etc.
Gopher- 1990/91



Univ. of Minnesota
Search for Data on campus
Spread
30
Before WWW (cont.)

WAIS - Wide Area Information
System



Thinking Machines Corp., Cambridge
Searched documents & made index of
words
All were short lived

But demonstrated what could be done
31
WWW - The Beginning



Doug Englebart: mouse + on-line
system, NLS
Vannevar Bush: 1945 paper - hypertext
Ted Nelson: Xanadu System




Computer Lib/Dream Machines
Hypertext: forms of writing which branch or
perform on request; they are best presented on
computer display screens
Worked on Xanadu during 70’s & 80’s
Apple Macintosh HyperCard - 1987
32
WWW Finally

Tim Berners-Lee @ CERN



European particle physics lab
Swiss- French border
Features and Goals



A shared information space, inclusion
Across platforms
URL- Uniform Resource Locator



To avoid database restrictions
HTTP- to replace FTP
HTML
33
WWW Early Years



Slow Start - few but CERN supported
Hard to program links
Just a few browsers Lynx & Viola
34
Mosaic

Marc Andreessen & Eric Bina






U. of Illinois
January 1993- released Mosaic, a
browser, over the Internet
Used Mouse, hypercard
Links in different color
Seamless integration of text and graphics
Re-written for Windows and Macintosh
35
Netscape Navigator

1994 – Jim Clark, Silicon Graphics


Univ. of Illinois – objected


Andreessen had been a student there
Clark & Andreessen



Commercialize Mosaic
Netscape Communications Corp
Mosaic died
1995 – Public release of stock

$28  $58 (day 1)  $150
36
Chapter 9
1981-1995
Workstations, UNIX
& the Net
37