% fortune -ae paul murphy

From Chapter Two: The appliance computing culture

This is the 18th excerpt from the second book in the Defen series: BIT: Business Information Technology: Foundations, Infrastructure, and Culture

Note that the section this is taken from, on the evolution of appliance computing, includes numerous illustrations and note tables omitted here.

The iSeries

Although DEC's VAX line both defined and dominated the mini-computer market for most of the eighties, the only significant mini-computer still offered commercially is the IBM iSeries.

Note, however, that many Unix machines from companies like Sun are used as mini-computers as are millions of Intel based machines running Unix variants like Linux or BSD. The difference is that the iSeries is intentionally sold only into this market while the gear used to run Unix in a "mini computer mode" is sold as a more generic machine which some users then treat as a traditional mini.

VAX - Virtual Address eXtension (on DEC's previous memory designs)

The VAX combined 32bit addressing with cheap floating point, vastly improved memory management, and a new proprietory OS called VMS - that eventually became the basis for Microsoft Windows NT 4.0. The first in the series, the VAX 780, was intended to match the performance of the base IBM 370/158 --a 1 MIPs mainframe. Later models ranged from the Microvax line introduced in 1985, to the largest multi-processors available from anyone in the late eighties. The earliest VAXes distinguished themselves by providing fast, accurate, floating point and thus achieved almost immediate success among researchers and academics who mostly ran Unix on them. The VAX was less of a success commercially until enough software had been generated in academia to make its adoption a necessity among engineering and related businesses. That, in turn created demand for other business applications on the VAX and let DEC become, by the late eighties, IBM's primary competitor among businesses without pre-existing data centers. DEC developed the first, and still fastest per cycle, 64bit microprocessor CPUs: the Alpha (shipped with Unix in June,1993) architecture, as the follow on to the VAX. Sadly, internal dissension about the role of VMS versus Unix eventually led the company to devalue Unix and thus its academic supporters, who reacted by moving en mass to Unix on Sun and later Intel. A subsequent attempt to emulate IBM by relying on services revenues associated with VMS licensing and support promptly failed because most commercial Vaxes ran packaged applications which didn't need support -- while new software development for the VAX essentially stopped when DEC reduced academic support for its Ultrix/OSF1 Unix offering. As a result DEC was eventually forced to choose between all-out support for Unix on Alpha in the academic and research market or business failure. In response it tried to deal with Microsoft in an attempt to keep VMS alive (as NT 4.0) but this failed and Compaq, (now HP) which bought the remnants in 1997, and has since end-of-lifed both VMS and Alpha.

The iSeries was introduced in 1979 as the System 38, but was unrelated to the System 34 before it or the System 36 introduced four years later.

Instead, the System 38 was a revival of the IBM Future Systems Project from 1968 -1972, which had been intended to produce a replacement for the 360 architecture but had then been cancelled in response to the extremely negative reaction of the System 360 customer base.

Thus:

• The original CPU ran at about 2.5MIPs (more than four times as fast as the 360 and significantly faster than most 370s when first shipped in 1979);

• It featured a microcode (meaning in firmware and therefore extremely fast) relational database that obsoleted both hierarchical and network databases;

• The original machine featured 48 bit addressing - giving it access to large memory spaces and the efficient task switching that the 360 completely lacked; and,

• It featured a highly interactive operating system environment (Control Program Facility or CPF) intended to support user database applications.

None of this, neither the hardware nor the applications model, fit what most IBM buyers knew and, as a result, data center managers who saw the prototypes in 1972 simply refused to have anything to do with it. What they wanted was a faster, cheaper, 360 - exactly what the 370 and its successors delivered -- not a machine that would turn their data centers upside down, invalidate most of their successes to date, and put users in charge of resource scheduling.

One of the oddest results of this was that the System 38, when finally introduced, was intentionally undersold by IBM, which characterized it as a "mid-range" processor for marketing purposes when, in reality, it has consistently been IBM's most advanced and powerful standalone system.

Today's version (late 2005) is sold as the iSeries and manufactured as a downgraded pSeries (IBM's Unix line) with custom firmware to run OS/400, V5R2; IBM's flagship mini-computer operating system. Together with hardware and services this system accounts for around $4.5 billion per year in IBM revenue. A typical single machine iSeries installation runs around a million dollars inclusive, while the machines currently range from about $4,000 to about $5,500,000.

Back in 1979, however, a wide range of other companies were making and selling highly successful mini-computers. Burroughs, Data General, Fujitsu, Harris, Hewlet-Packard, Honeywell, ICL, Nixdorf, Prime, Wang and many others built their initial reputations on machines sold with specific applications and then branched out to sell the same gear for more general purpose use.

Prior to the VAX, none of these were commercially important to IBM, which didn't see them as competitive and did see some strategic value in keeping them alive in view of past government anti-trust action against it.

RDBMS (relational database management system) - the 60 second primer

Database theory is one of the most complex subjects in computer science with almost all of today's significant business applications ultimately dependent on a relational database management system. Luckily most auditors, managers, and business owners do not need to know much about the internals (but see Chapter 5). Key things to remember include: A database is a collection of data together with an access method for getting at that data. Oracle is not a database, it is a company or a product. A phone book is a database of information linking names to phone numbers and accessed using a KSAM - keyed sequential access method- index. The theory behind relational databases comes from set theory (known as the theory of relations in formal English). You may recall seeing Venn diagrams in school - that's set theory at work. On most machines a database management system fits between the hardware operating system and one or more applications which use the data. Its fundamental purpose is to isolate the data from application error while reducing duplicated data storage and error by allowing multiple applications to access the same information. The software involved is usually called a relational, or object relational, database management system (ORDBMS or RDBMS) and handles data storage, access authorization and logging, transaction serialization, data retrieval, and data backup and recovery. A relational database stores data according to a set of rules in files, known as tables or relations, consisting of rows of data. Each row, also known as a tuple, must contain both a key that uniquely identifies the row and zero or more columns of information. Rows must be structured according to Boyce-Codd normal (meaning with duplicated information removed) form. In practice many business databases are "denormalized;" with processing efficiency usually cited as the reason. In general this is not true, denormalized tables are simply easier for COBOL programmers to work with, but this usually lead to inefficient code and/or poor business processes. SQL (structured query language) is the usual tool for manipulating data in a relational database system but is not limited to relational systems. There are, for example, SQL front ends to IMS.

The VAX changed that in part because it offered about the same performance as the 370/158 at a quarter of the cost, but mostly because DEC announced its intention of using the VAX to break out of its traditional floating point and scientific processing markets and into IBM's highly profitable small and mid range business markets. DEC didn't actually have the software for this, but some people within IBM saw DEC's announcement as a lever to finally get the server piece of the abandoned Future System out the door and into the market as the System 38.

The System 38 lineup was an enormous technical success that hardly anyone recognized and IBM sold relatively few units until it was revised and re-released as the AS/400 in 1988. What made it such a technical success was the fact that it was designed, from the ground up, to act as a highly reliable interactive processor.

Unfortunately, the database driven interactive applications model was also why the original System 38 sold slowly. People, particularly those with data processing experience and thus most likely to influence purchase decisions, either saw it as a threat or simply failed to understand it.

Some notes:

1. These excerpts don't (usually) include footnotes and most illustrations have been dropped as simply too hard to insert correctly. (The wordpress html "editor" as used here enables a limited html subset and is implemented to force frustrations like the CPM line delimiters from MS-DOS).

2. The feedback I'm looking for is what you guys do best: call me on mistakes, add thoughts/corrections on stuff I've missed or gotten wrong, and generally help make the thing better.

   Notice that getting the facts right is particularly important for BIT - and that the length of the thing plus the complexity of the terminology and ideas introduced suggest that any explanatory anecdotes anyone may want to contribute could be valuable.

3. When I make changes suggested in the comments, I make those changes only in the original, not in the excerpts reproduced here.