% fortune -ae paul murphy

Costs Relative to Client-Server

This section introduces the three cost scenarios to be used throughout the text and explains the idea that a big organization, like an FAA or a Fortune 5,000 company, consists of islands of IT opportunity that can't simply be forced into a single mold but have to be transitioned at different rates and different times depending on the people in place at the beginning.

Thus the example scenarios involve:

1. a 500 user faculty system;

2. a 5,000 user manufacturing company with three major sites; and,

3. a 30,000 user international professional firm with about 500 offices.

Each cost comparison looks at cost issues relative to the Microsoft client-server alternative with respect to capital, maintenance, support, networking, telecom, HVAC, and business risk.

In addition each scenario will be discussed with respect to:

• Minimum sensible networking options including telecom integration and appropriate administration tools;

• Server sizing - AMD, USIV+, and T1 now; USV/APL and T2 next year.

• Specifics of the Sun Ray 2[FS] and 270 displays. Resolution, power use, noise, reliability, wiring options, theft/vandelism issues and the importance of screen sizes and types with respect to readability, comprehension, and retention;

• IT staffing and organizational charts for both the thin client stage and the full smart display architectures;

  The emphasis here is on the staffing consequences of cross training and job expansion with the smart display approach. The key difference is that the traditional IT organization is largely about what staffers cannot do: they have roles they cannot be allowed to go beyond. Thus someone with Solaris administration, RDBMS, and networking skills will be frustrated every day by the limitations imposed by the job description.

  The smart approach reverses this: using cross training within teams and ad hoc role assignments to ensure that all available skills are tapped for the common goal - with two key results:

  • greatly reduced turnover; and,

  • the virtual elimination of mistakes and omissions.

• Organizational change - change processes for the pre-existing wintel organization as it adapts first to thin clients and then to smart displays.

• Key implementation issues
  • for a 500 user academic system in a single location:
    • Windows support for visiting and change resistant faculty and students;

    • Windows support for Wingot proxy licensed services (like Lexis/Nexis access);

    • Home user integration, xDSL opportunities, and off-campus computing;

    • Campus wiring (high threat ==> use optical where possible)

    • Faculty or group specific zones with assigned primary/secondary admin responsibility for each one;

    • DHCP/wireless access issues;

    • Unusual functionality comparisons -e.g. on the fly compilation in classroom and session continuation across locations.

  • for a 5,000 user integrated manufacturing business with three major locations and a half dozen marketing offices:
    • no down time tolerable ==> redundant data centers (one at each site) and cross trained teams.

    • no down time tolerable ===> use of N1 to map resources to demands;

    • SOX auditors trying to re-impose organizational controls from the 1920s;

    • conflicting user agendas can lead to IT working against itself: solution is to use Solaris zones to isolate groups if and where necessary, support significant process variation between groups but maintain database consistency.

    • data center distribution, cross training requirements, and open source support as the answer to the odd problem that the better the staff get, the less work they'll have to do.

    • balance production users vs wintel support in finance (may require additional staff);

    • assume Oracle financials/ERP for costing and sizing purposes; assume 10% custom screens added internally over first three years.

  • for a 30,000 user world-wide professional services firm with 500 offices:
    • system sizing and layout: the company is big, the offices are small; but national groupings face locally unique professional practice and data control issues. Hence mix of small and large servers; rigid client/firm data differentiation; mix of Sun Ray device types to support portability; mix of data centers to support service continuity without forcing data across jurisdictional boundaries.

    • Extensive Wintel integration issues: no wireless, some DHCP, rigorous document and data security - emphasis on Sun Ray physical security for data - stolen devices become a non issue.

    • Extensive remote access issues (from client sites and other offices) including legal complications related to allowable data movement and the imposition of local procedural rules on remotely accessed data. E.g. Access from Canada to data in Germany proceeds under EU data access rules, not Canadian ones.

  • Draw summary conclusions applicable to smart display vs client-server in all three groups -look particularly at user productivity, risk-to-data issues (transfer of responsibility for meeting both legislated and actual responsibilities for client data from professional staff to system).

  Bottom line for thin client: cheaper, safer, faster.

  ---

  Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.