'

From Chapter Four: The Unix and Open source Culture

The Unix culture is that of the graduate student in a research function, not that of the bureaucrat in a data processing role.

About this grouping (3)

There is a unique Unix culture, but it developed and flourishes in the research, not business, community. In the research community cost and performance pressures on individuals combined with the general absence of cross cultural contamination to produce Unix best practices like peer review, networking, user empowerment, the open source movement, and direct end user control of systems decision making.

Those management ideas have only recently started to appear in business uses of the technology as open source products have started to displace proprietary solutions.

In large part this is due to the head start other toolsets had in developing business oriented computer cultures. Large companies generally select systems managers using their experience in other large companies as a proxy for expertise, thereby failing to recognize that the skills needed to properly deploy a technology tend to reflect that technology. As a result most of the executives responsible for large Unix sites have mainframe, mini-computer, or Microsoft PC backgrounds that lead them to mis-manage Unix.

Nevertheless the Unix cost and performance advantage is so substantial that even mis-managed and under appreciated systems often become critical to corporate operations. For example many large companies are unware that they use Unix at all but:

  1. Most have at least some Unix based server, storage, firewall or other special purpose appliances;
  2. Most use Linux or BSD based PC servers for web and email access;
  3. Most larger organizations have one or more large Unix machines in use as database servers in otherwise fully Microsoft dominated client-server architectures; and,
  4. Many organizations have one or more Unix machines, often quite old ones, in their otherwise mainframe oriented "shops" -either as experiments in cost reduction or to handle tasks for which mainframe software was unavailable, excessively impractical, or simply too expensive.

To the extent that a "true Unix" architecture exists in business and administrative organizations it is characterized by highly inter-connected servers with either smart displays like this NCD or powerful engineering and graphics workstations as desktops.

When this structure is successfully ported to business environments the resulting systems architecture strongly resembles that of the original application appliance group - both use smart display desktops and central processors - and is its logical successor in terms of management methods, user focus, and required controls.

In the iSeries or mini-computer world, IBM's artificial perpetuation of the high costs from the seventies and early eighties means that control is now highly centralized with IT staff members having little or no freedom to respond immediately to day to day changes in user requirements. Positions are intensely hierarchal; budgeting is usually annual and "cast in concrete;" and, there is a general focus on stable delivery of core functional support applications. Fundamentally IT staff work for the IT department.

In a well run Unix environment processing is centralized but costs reflect current technologies and are small enough to allow actual control to be distributed. IT staff are encouraged to adapt processing or resource allocations to user needs and annual budgets set flexible boundaries, not hard limits. Fundamentally IT staff work for users and user departments.

Unix hardware costs less, often much less, than alternatives. Obviously any machine that can run a licensed Windows OS can run a free Unix, but cost comparisons for bigger gear are far more dramatic. For example a high end iSeries can cost in excess of $4.5 million but a comparable Sun Starfire 12K runs well under two million. That contrast is even more pronounced in the mainframe world where a 72 CPU Sun 25K at just about two million dollars will easily outperform a 128 core mainframe at well over twenty-six million - and that's before the mainframe's additional software licensing and staffing costs, which don't exist for Solaris, are considered.

The result is a contrast between a Unix culture in which processing power is considered cheap and plentiful versus one in which processing power is scarce and expensive. Properly implemented, a mainframe data center focusses on maximizing the use of expensive resources, but a properly managed Unix data center does opposite - it pushes control and processing power out to the users because that's both cheaper and more effective than imposing usage controls.

Costs and processing power are important but, from a corporate deployment perspective, the most significant benefit Unix offers is not directly related to initial capital cost or processing speed. What is most significant about Unix is that its combination of reliability and continuity allows the organization to side step both the product churn characterizing the Microsoft PC environment and the financially enforced rigidity of the mainframe world.


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.