Table of Contents

• Calendar of Events
• Next General Meeting -- Apr 14
  • Tutorial: To be announced
          by someone special
  • Presentation: The OSI Model and its role in explaining communication between networks
          by John House
  • Book Specials
• Next Linux Meeting -- Apr 15
  • Presentation: Novell commitment to cross-platform environments
          by Evan Bills
• Meeting Locations
• Upcoming Events
  • Manager of Manager Design Principles
  • Scope analysis, the benefit of a clear purpose and maximum freedom
• Sponsors
• Special Interest Groups (SIGs)
• Volunteering
• Contacts
• Submitting Articles

Next General Meeting is Wednesday, Apr 14

Tutorial

We have two possibilities for tutorials this month, but neither is confirmed at this time

Any member of the St. Louis Unix User Group or an affiliated organization is welcome to present a tutorial. SLUUG members are an informed, but supportive audience, and we appreciate it when you share experience and insights, or demonstrate software and utilities that could help us.

Bio:

Please contact Terry Linhardt if you would like to make a tutorial presentation. www.sluug.org/contacts/

Presentation

Abstract:

The Open Standards Institute (OSI) model describes the roles of hardware and software components in a network. It defines strata of media and functions, and explains the interaction between them. While this model is largely defunct as a standard, it is used universally to define the relationships among elements of a protocol stack and to compare different protocol stacks.

Tonight's presentation offers general insights into data transmission between networks. The OSI model is the focus for definition and abstraction of the roles of hardware and software elements within this communication.

Bio:

John House runs SUSE, openBSD and freeBSD hosts at his home. He is a willing student. He is also willing to share information as soon as he learns it.

John is also responsible for the typos in the CRONicle. As editor, he strives to diminish that responsibility and to increase the number and variety of interesting articles and commentaries on technology issues.

Ideas, questions and suggestions are welcome; please contact Terry Linhardt (presentations@sluug.org).

Book Sales

(booksales@sluug.org).

Next St. Louis LUG Meeting Is Apr 15

Novell has a long history supporting cross-platform environments, and with the recent acquisitions of Ximian, the creator of the Outlook replacement Evolution, and SUSE, a major vendor of Linux, Novell has committed to supporting the Linux community. Evan will present Novell's strengths in supporting and integrating these cross-platform environments, and provide information on the value Novell expects to bring to the Linux community.

The acquisition of SUSE will add a significant amount of combined value to the solutions Novell can bring to the Enterprise marketplace, and we will focus on Novell's products and technologies available for the Enterprise today.

Evan Bills is a Central Network Specialist and a Linux Specialist for Novell. Although he joined Novell recently, he has had 12 years' IT experience, mostly with Linux. In his own words, "I am very excited about the moves Novell is making in the Linux industry and hope to share that with your group."

Comments, questions, and ideas for the St. Louis Linux Users Group are welcome; please send email to linux@sluug.org.

Meeting Locations

SLUUG - St. Louis UNIX Users Group General Meeting (http://www.sluug.org)

SLUUG - St. Louis UNIX Users Group Steering Committee (http://www.sluug.org)

St. Louis Linux Users Group (STLLUG) (http://www.stllinux.org)

St. Charles LUG (http://www.sluug.org/~stclug)

Security UG

Hazelwood LUG (http://www.sluug.org/~hzlug)

CWE-LUG - Central West End LUG (http://www.cwelug.org)

Advanced LUG (http://stladvlug.org/)

Extreme Programming (http://xpstl.org/)

MOSLUG - MO Open Source LUG (http://www.moslug.org/)

STLBSD - St. Louis BSD Users (http://www.stlbsd.org)

Group is planning a new meeting schedule. Consult website for current meeting information.

St. Louis Java Users Group (http://www.ociweb.com/javasig/)

Gateway JUG (Java Users Group) (http://www.gatewayjug.org/)

STLWEBDEV - St. Louis Web Developers (http://www.stlwebdev.org)

GAMUG - Gateway Area Macintosh Users Group (http://www.gamug.org)

SILUG - Southern Illinois LUG (http://www.silug.org)
O'Fallon meeting

CORCC - County Older Residents Computer Club (http://www.a-zuc.com/corcc/)

Meets (almost) every Friday from 1:00pm to 3:00pm. Provides free classes for seniors 55 or older. For info, contact Tom Kimber (t.kimber@worldnet.att.net) or Arthur Hollman (ajhollman@juno.com).

Prairie Commons Library
915 Utz Lane (between Howdershell and Dunn)
Hazelwood, MO

We publish meeting schedules of groups in the St. Louis area that may be of interest to our members. If you would like to have info added about your group, please mail the newsletter editor whose link is on sluug.org.

WARNING: These articles and notices may express personal opinions and SLUUG exerts no more editorial control over such content than does a public library, bookstore, or newsstand. Any opinions, advice, statements, services, offers, or other information or content expressed herein are those of the respective authors and not necessarily supported by SLUUG. SLUUG does not guarantee the accuracy, completeness, or usefulness of any content, nor its merchantability or fitness for any particular purpose.

Manager of Manager Design Principles

Anyone can implement a Manager of Manager (MOM) in some fashion by contacting a reliable vendors. Vendors will predictably recommend replacing many of your processes with their proprietary processes and your hardware with their proprietary hardware. Vendors will typically support their solutions using their equipment as you adjust your needs to accommodate this. Although this viewpoint is cynical, it often reflects reality.

Our goal is to generalize the network administration and system administration practices some enterprises use to minimize the disruption to their network design and administration. This goal has many consequences:

• Your responsibility and accountability as system administrator or system analyst increase.
• You give your team the opportunity to fundamentally determine network design, function, and performance.
• Your internal clientele (coworkers, subordinates, and leaders) have greater incentive to cooperate for two reasons: you have the potential to offer less interference than a vendor-specific solution and your clientele can have ownership in the network's characteristics and performance.
• Your organization has greater flexibility and productivity.
• Your organization has employed flexible intellectual resources rather than rigid capital resources that carry long-term cost and significant risk of obsolescence.

A prudently designed MOM needs vendor input to provide multi-site expertise, well-written and debugged software, and experience with a variety of hardware. But you need to ask who will "drive the bus." One approach is to use a "best of breed" selection criteria and to integrate optimal contributions from one or more vendors. Such thinking is the antithesis of a single-source vendor selection process.

In MOM design, the first reward for taking responsibility is freedom. Predictably, the finished product of a MOM design will be hierarchical, like a Cascading Style Sheets. Processes will send status messages to monitors, which in turn summarize and pass data upstream to other monitors. However, your design of the MOM is more analogous to creating a key-bound database. You have the freedom and responsibility to arrange granular and disparate information to identify critical functions and describe the condition of the entire enterprise's IT infrastructure.

A MOM reporting system can describe the enterprise in any of the following respects:

• Monitoring of physical media and hardware. Correct monitoring encompasses both assets owned by the enterprise (such as servers, routers, and LAN segments) and those which are often leased (such as T1, T3, OC3, OC12 circuits, POP routers, and hosting servers).
• Monitoring of nodes, hosts, and processes in the network design
• Monitoring of processes relevant to business function, and independent of the locations of hosts on which they reside. This aspect of monitoring watches the inventory system, the database, the client/server application that links the database to the end user, and perhaps the financial application that abstracts reports and graphics of cost and revenue which flow from the inventory function.
• Monitoring a network by layers within the OSI model [The April 14 SLUUG presentation addresses the OSI model]. This concept can give a concise picture of an event within the OSI stack and expedite troubleshooting.
• Redundancy can be incorporated both within the above-listed monitoring designs, and among them, to provide fault tolerance and multi-dimensional presentations of enterprise's IT condition.
• Logging to provide the alarms and meaningful data to describe them.

You are correct in noticing that above-listed bullets are redundant. Good MOM design is concise because it continuously provides views of the enterprise through overlapping sensors. Good MOM design distinguishes a server failure (hardware) with a process failure (program or stack). You will know if an event is a failure of one process on one server or an enterprise-wide application failure on hundreds of hosts, for example.

You can choose to design the MOM in a manner that is concurrently hierarchical and decentralized. Your enterprise's departments and business operations have had monitoring for their own areas of responsibility. For example, a NOC will monitor circuits while a database group will monitor database processes throughout the enterprise. A decentralized approach permits groups to continue using their processes, while you use the traps that are currently in use to be passed to an enterprise-wide MOM. Therefore if a new business group determines that a novel monitoring system is optimal for its purposes, this model of MOM development would more likely to incorporate that new concept.

Such flexibility requires monitoring software that is highly interoperable. The vendor providing such software must support interoperation with a wide variety of platforms, processes and hardware. Scripting languages become very important: both the IT staff and the vendor need to become adapt at script and API design. Such skill includes the craft of programming and the art of learning new programming or scripting languages efficiently. Proficiency with open protocols such as ICMP, SNMP and MIB enable interoperability based on industry standards and butress your vendor independent approach.

Because a MOM oversees an enterprise, it necessarily aggregates the data and generalizes it. Several characteristics follow from performing this generalization properly:

• Generalization is essential. It is confusing to represent the failure of a network segment as a failure of 100 servers.
• Generalization must allow for "drilling" into a given trap for meaningful and granular detail. A well designed trap will distinguish between a media failure on a switch, an OSPF storm, or a failure of a network segment.
• Good MOM design provides a great deal of information concisely. For example, a UPS alarm, a fire alarm, and a down circuit alarm at the same site combine for a highly diagnostic picture. A process failure in the absence of segment faults or network problems is also diagnostic.
• Good MOM design provides redundancy in alarms and monitors. The failure of component media or hardware does not blind the MOM if it is well designed.
• Alarms need to inform people who can fix the problem quickly. However, correctly designed alarms do not disrupt people who have no contribution to the problem.
• Persons who can fix problems should own the trouble tickets. These tickets become the primary basis for informing monitoring staff and persons with an interest in the problem of the progress. A trouble ticket is part of the communication in the troubleshooting process, not a clerical chore.
• Like a good alarm, a good trouble ticket provides a general summary of the problem and the current status. It also provides a means to "drill" into the generality to determine the specifics of the problem and the paths for resolution.

Moving to a MOM (Manager of Manager) solution offers myriad potentials for creative and responsible IT planning. Rather than becoming technical experts supporting a procurement decision, you can chose to maintain the organization's technical investment, conserve infrastructure value, and meet business needs. IT managers should continue to rely upon their craft and imagination to meet enterprise needs in the most flexible and cost effective manner.

Scope analysis, the benefit of a clear purpose and maximum freedom

On March 26, Ivan Sutherland spoke at Washington University on "The Importance of SCOPE in software, hardware, design, economics, and life." Dr. Sutherland is a Sun Fellow and a Vice President at Sun Microsystems.

I cannot do justice to the topic, but Dr. Sutherland's remarks left me with something you might find useful. A well stated scope makes design happen and avoids legacy (aka baggage or structural obsolescence). Specifically, a well defined scope limits the parameters of the problem, assures maximum freedom in generating solutions, and assures interoperability.

Let's say that you and three companions are on vacation in a foreign city, and you want to find a landmark that is not published on a tourist map. You know it is nearby, but guides have no knowledge of the landmark. So, each companion sets off on an agreed purpose: to find the landmark. You have also agreed that each person is to stay in touch by cell phone, and no one should travel in the same general direction.

All of your companions go in different directions, asking residents as best they can about the landmark. Eventually, each person finds it, but one of your companions knows the most direct route. Another knows a route that passes the best restaurants and shops, a third companion knows a route that also leads past other sites of historical significance to this exotic country, and the fourth, well, he found a watering hole with great music that was known only to local people, and that is where all of you spend the evening.

By defining a simple, clear scope, then leaving people freedom to pursue it, you have several ways to reach your purpose. Then you can choose the best way for the moment's needs (optimization).

Sponsors

• Accenture (formerly Andersen Consulting) - http://www.accenture.com
• Ajilon - http://www.ajilon.com
• Analysts International - http://www.analysts.com
• Bradford & Galt - http://www.bradfordandgalt.com
• CIBER, Inc. - http://www.ciber.com
• Cybercon - http://www.cybercon.com
• Cypress Systems, Ltd. - http://www.cypress-systems.com
• Daugherty Systems - http://www.daugherty.com
• Graybar Electric - http://www.graybar.com
• IBM - http://www.ibm.com
• Informix Corporation - http://www.informix.com
• K&S Pritchard Enterprises, Inc. - http://www.kspei.com
• Maryville Data Systems, Inc. - http://www.maryville.com
• Maverick Tube Corporation - http://www.maverick-tube.com
• O'Reilly & Associates - http://www.ora.com
• Renaissance Worldwide, Inc. - http://www.rens.com
• Sun Microsystems - http://www.sun.com
• Stopka & Associates - http://www.stopka.com
• Sunnen Products - http://www.sunnen.com
• Sykes Enterprises, Inc. - http://www.sykes.com
• Technisource Inc. - http://www.tsrc.net
• Triple-I Corporation - http://www.triplei.com
• Venmar Systems, Inc. - http://www.venmar.com
• The Law Offices of Rich McLennan - mclennan@sluug.org

For more information about sponsoring the St. Louis UNIX Users Group, contact Christine Wanta.).

Special Interest Groups (SIGs)

Volunteering

Contacts

To relieve our officers and volunteers of unwanted SPAM, we no longer include email addresses in the newsletter. Nonetheless, we are eager to hear from you. You can find links to our email accounts on www.sluug.org/contacts/

Submitting Articles to the SLUUG Cronicle