INFS2052 lecture 5.1

Standards and Open Systems

References

Tanenbaum section 1.7
World Wide Web Consortium web pages

Data Communications Tutorial on Standards

Standards in general, and in computer communications

General examples from audio cassettes (physical, recording standards (mono/stereo/quad, simple analogue, Dolby, digital), audio connectors; electrical power supply voltages, AC/DC, frequency, connectors; telephone dialling/signalling, sound encoding - voltages/currents, frequencies

Computer and data communications examples from Morse, ASCII, modems, FTP, HTML, programming languages, operating systems

A Standard is a way of recording a common ground definition of some system at some (extended) moment in time. It is contained in a specification or an implementation.

Standards may be

• proprietary
• public - open

and may also be

• formal
• informal

Why have standards? pro

In many commercial areas:

it is thought to be desirable to to have a larger market available to all vendor's products (an efficient market rather than locked in, non-overlapping, small groups of consumers)

In software and hardware - allow efficient porting of programs, systems and/or correct, consistent results of running a program on multiple systems. This is testable... test cases, compare results.

In communications - standards are essential for interoperability. Hard to test in general: deadlock, crash, lost data, sloooow communication...

but see con later

Development process of standards

Types of standards - de facto and de jure

Some products, processes etc. become so widely adopted in some proprietary or evolved form that they become a de faco standard.

examples

• IBM PC/DOS operating system
• Unix operating system
• original Pascal and BASIC programming languages
• original SQL query language

In many cases a wide variety of forms and variants are proposed or in use. To get a common standard

• the vendors may combine in a consortium to attempt to reach consensus;
• researchers/experts and vendors may be brought together by government;
• researchers may bring themselves together

Examples

• standards adapted and extended from existing pre-existing technology as a defining example - e.g. Pascal, BASIC
• standards set by writing broader specification from lightly documented existing implementation - e.g. TCP/IP, email (RFC822)
• standards created as design development to take the best features from and extend a wide range of existing implementations or ideas- e.g. COBOL, OSI, Ada [1983]
• standards written to merge and accept/reject new developments of existing standards - e.g.
• HTML 3.2, HTML 4.0
• HTTP 1.1
• Fortran 90 - from FORTAN 77, from FORTRAN (alias FORTRAN 66 or FORTRAN IV) from IBM FORTRAN [1958] and at least 7 other versions.
• Ada 95

HTML went through several versions very quickly, defined by different bodies:

• HTML 0 -   original CERN specification (multi-government funded nuclear physics research laboratory), text only approx 1991
• HTML 1.0 - added images - defined in Mosaic implementation (NCSA - government-funded US research lab, Feb 1993)
• HTML 2.0 - common usage approx June 94, formally defined in RFC 1866 (IAB - Internet Activities Board) Nov 1995
• HTML 3.0 - March 1995 - a testbed implementation existed but the language was found to be too unwieldy, abandoned
• HTML 3.2 May 1996
• HTML 4.0 draft July 1997

The formal process

In ITU and ISO "working groups" or "study groups" of 10s to 100s of expert individuals representing nations or corporate entities volunteer to meet on specific topics suggested by members, approved by a standing committee structure.

In ISO a standard goes through 3 stages:

1. committee draft
2. draft standard
3. standard

An international working group will meet two to four times a year to create a Committee Draft, by consensus.

The CD is sent to all National bodies (who have standing or special working group committees - in Australia, from universities, CSIRO, computer vendors) for comments and vote whether to accept at this stage - from which the working group/technical committee creates a Draft International Standard DIS by consensus. Comments are openlky puvblished to all member bodies, and must be explicitly addressed and replied to openly by the working group.

The DIS is likewise voted on and comments incorporated by the working group into the IS.

Similar processes occur within national bodies, particularly in the USA ANSI, IEEE etc.

The consensus process means that an uncooperative member can block standards or widen them to the point of no interoperability.

Standardisation bodies

ITU - telecommunications

consists of PTTs

- largely (until now) owned by or part of governments, represented by public servants (experts)

handles international matters since 1856 e.g. telegraphy - Morse code...

radio spectrum - frequency allocation

geo-stationary satellite positions and frequencies (?)

modem protocols

RS-232 - and connectors

ISO - International Organisation for Standardisation

umbrella body of National Standards bodies, 1 per country;

national bodies are (in UK, USA, Australia) government supported but commercial membership

e.g. Standards Australia, British Standards Institute (BSI), American National Standards Institute (ANSI)

IEEE

Institute of Electrical and Electronic Engineers (USA)

an engineers' professional body but very concenred with educational standards and operational standards for a wide variety of items, protocols, and processes

e.g. Ethernet standards were originally IEEE defined, then made international by ISO; C, C++ and POSIX (standard UNIX) likewise

ANSI - American National Standards Institute

industrial collaboration - where the ANSI standards including ASCII character set and Fortran come from

Internet Architecture Board

originally researchers, a few vendors - now mainly vendors (long term interests, stability...)

Industry etc consortia: e.g. World Wide Web consortium W3

universities and software developers/vendors concerned with WWW, serviced by MIT and INRIA.

Interaction

individual members of both bring cross-fertilisation

official - e.g. IAB RFCs include HTML, HTTP specifications

- to get wider dissemination and approval for W3C proposals

Why have standards? con

standards wars - standardisation as a weapon

• Microsoft, Compaq and the browser icon on licensed versions of Windows 97 (Netscape Navigator or Microsoft Explorer
• Microsoft, Sun and Java 1.1 - ditto

limitations - interoperability may not be ensured by published standard alone - OSI example

moving targets

• HTML moves too fast for its own developers!
• but compare Fortran process in long term (a revised standard about once in 10 years (1966, 1977, 1990,...) with warnings of "deprecated features" which were OK in the past, still present in standard, but will be removed from the next standard - 10 year warning!
• "no standard can be finalised until the technology is static, i.e. dead"

Non-working standards processes in some cases e.g.

• Algol 68 (specification a major study to understand, very rarely implemented, no longer in use)
• OSI (moot for 10 years, not completely implemented, never displaced Internet standards)
• C++ ((continuing tussle to define an acceptable standard - too big, too many conflicting demands on the standard)

Conclusions

Standards are essential in data and computer communications

To accomodate the speed of change requires lightweight standardisation processes and consensus/development through Open Systems.

An Open System is one with "an architecture that uses standards published for public use so that compatibility is achieved between diferent users and manufacturers of peripheral products and software" [J.C. Nader, Prentice-Hall's Illustrated Dictionary of Computing, open architecture]