|
Professional Experiential
Portfolio
for
J. William
Bennett
|
Artifacts protected by Password:
Some artifacts herein are password protected due to
methodology content. For access to these files contact
mail@JWBennett.info and I
will supply passwords to those with confirmed interest. |
|
Systems development is a core competency. I
have been actively engaged in programming for over 20 years. I have
authored both commercial software systems and industry application
solutions. While much focus is often given to my commercial
systems experience, I am more energized by solving the smaller and more
finite issues facing clients and employers alike.
- DERIS:
Chevron Corporation's General Manager of
Global Shared Services was tired of going into meetings and being
'blind sided' with questions like 'When will Brazil's link be back
up?' (he never knew it was down!). He often has no information on
the status or resolution of an incident. Root cause analysis takes
20-60 days. The problem is exacerbated by multiple silos of
accountability and responsibility among the 3500 locations in
Chevron. Some services are outsourced, some run by business units,
some by central IT. What is happing, where its happening and what
are the impacts associated are nearly impossible to ascertain. An
ongoing central deployment of ITSM cannot bridge many of the silos
in question.
I was contracted to develop a solution
and manage the resulting project: I lead the development of the
Downstream Enterprise IT Reliability Information System (DERIS).
This system was designed to provide and end-to-end "critical
business process" view of reliability.
I started by directing the project team
to define the critical 'business sensitive' transactions (e.g. those
transactions that represent customer facing impacts to Chevron
revenue or customer commitment) and the key revenue generating
locations in which they run. I then designed and directed the
development of a reliability database, a browser based dashboard and
a set of synthetic transactions (run remotely) to report the
reliability and performance of these transactions from the end-user
perspective.
In it's first release DERIS is giving
Chevron a view of critical transaction reliability (performance,
availability and accuracy) - by business process, by location, by
business unit - in a standard SharePoint dashboard. The next phase
will be to connect this capability to ITSM and RCA projects for an
integrated system. Artifacts from this development are
protected by NDA with Chevron, but may be disclosed under mutual
agreement between Chevron and the interested party.
-
OPCON/XPS:
In 2000 I
was approached by the president of Software Management Associates (SMA) in
Humble, TX. to develop a set of IBM mainframe agents and functions that
would allow their client/server based scheduling system to schedule events and
batch jobs on a modern IBM mainframe.
The agent was code named OpCon/xps390.
It was designed to be one of the more sophisticated scheduling agents ever
designed for large Mainframe operating systems (OS390 & Z/OS). Unlike JOBTRAC, I
retain joint copyright to this work and have included both sample
source code and
technical description of this effort.
This project was an initial foray into Web Services and Services Oriented
Architecture (SOA). It required that an RDBMS manage
services to agents on an Z/OS Sysplex from UNIX, UNISYS, Linux and Wintel
Servers. This was accomplished with an early WSDL-like XML construct.
The competed project involved over10,000 lines of IBM ALC code for the
mainframe, a sizeable TCP/IP internal communication driver development for
Z/OS, Visual Basic applications for JES2 access from a Wintel client, C++,
web development, HMTL, XML and SQL Server [OBDC] development. I also
configured, installed and supported the IBM mainframe development lab for
the this project (MP3000, Z/OS, LPAR(3), Sysplex). This project
completed in 2003 with Version 3 of the agent.
- JOBTRAC:
While contracting for Cameron Iron
Works, a major oil company equipment supplier in Houston [1982], I
began the development of a production batch scheduling system.
This was more than a utility, it was a complete package of
enterprise
tools and automated processes controlled by rules and calendars. The
product was called JOBTRAC and I successfully marketed JOBTRAC to 35
clients from 1985 to 1988 [see also "Marketing"]
The product was acquired by Goal
Systems [thru acquisition, now Computer Associates "CA"], and is
still being marketed today, nearly 20 years after its original
release [see
http://www.ca.com/us/products/product.aspx?id=1334
].
Although the product has been updated with client/server integration
and interfaces to other products, its internal workings are still
90% the same as in 1988.
Artifacts such as program code and
copyrighted manuals are not available due to contractual agreements
with the successor in interest to this intellectual property. I did
however poll a mainframe user news organization
(e.g. IBM Mainframe Discussion List [IBM-MAIN@BAMA.UA.EDU]) looking
for JOBTRAC feedback. Results of that inquiry are attached
hereto.
-
Tools, Utilities and Interfaces:
Besides the major developments of JOBTRAC
and OPCON/XPS which included hundreds of thousands of lines of code
and years of development and support, I have written dozens of
smaller systems on many different platforms. Clearly the scope
and complexity of a project is not a measure of its value.
Often it is the ability to 'see' the problem in a different way,
discern a problem from a symptom and
develop a solution.
In this list of accomplishments, I will
list first the 'problem' then the 'solution' as designed.
Problem: Verizon IT (GTEDS) has
several regional data centers that report computer usage by task id.
Applications programmers work on projects for over 30 states
telephone companies and the time charged to each company by the
GTEDS subsidiary is controlled by task id. There are anywhere
from 10,000 to 15,000 task ids in effect nationally at any time.
Task ids are part of the accounting code used on JCL and TSO logons.
They are assigned manually and a group of 5 people in the Tampa, FL
headquarters of GTEDS reconciles all the task ids from all data
centers every month. They discover and reassign about 4,000 task id
errors each month.
Solution: I wrote an internal SVC exit
routine (much like a modern day dll, SCI or driver) that loads a table of
valid task ids into memory. This SVC is dynamically activated,
deactivated and refreshable. I then set up calls to this SVC by all
JCL and TSO entry points. In 3 months error rates go to zero.
Five people get better jobs. Billing to state phone companies
is done 45 days sooner and is accurate.
Problem: Interfirst Bank in
Texas has a problem with applications programmers putting new
programs into production before they are fully tested. Sometimes
they are true emergency fixes and test time is not available, but
often it's just lack of discipline. One of the biggest problems is
an audit trail. There is no positive identification of who did what,
when or why.
Solution: Over 18 months in
development, the Automated Change Management (ACM) application
provided 3 interactive levels of authorization for changes to be
applied. I was the sole author of the system. ACM included a front end to IBM's Linkage Editor
build program
that logged all program update activity to production bin libraries. A
central program task was created that applied changes 24hrs a day. A
complete log of all changes and their attributes was maintained.
Changes could be scheduled or sequenced (don't apply this change
until after this job runs or this other change is made, etc.). These
features are now standard in many of today's systems, but the user
friendly nature of ACM actually made it easier to apply changes than
the old process of running your own job. That's still a problem in
today's systems.
Problem: I'm engaged in my
first big consulting contract at Cameron Iron Works (CIW) in
Houston. They have an accountant as a CIO. He can barely spell CIO,
but he is an extremely good accountant. He wants to associate the
technical service level objectives (SLO) to something he can understand -
financial impact.
Solution: I developed a reporting tool that
gathers information from computer task monitors, user logon
activity, disk seek data, tape mount data, etc. and compile a report
that associates cost of processing with services. Then produce
data on response time and report delivery service. This data
and reporting was successfully used to upgrade DASD and add memory
to processors by demonstrating the cost of NOT doing so.
Problem: Sysco Corp help desk keeps
responding to the same 'known problem'
over and over. There is no integration between the logging
system used to log problems and the technical support people who
work on the problems. No priorities no escalation.. a mess.
They are an austere organization. No money will be spent on fancy
problem tracking software.
Solution: The only system available to
both the help desk and technical support groups was the IBM VM system used
for email. Using VM/CMS and ISPF for CMS I developed a CMS Dialog
manager using ISPF tables to log incidents and assign
responsibility. An on-line reporting system was developed in
VM/CMS to report on status. The status report was made available
corporate wide and reduced the status request calls to the help desk
by 80%. The new system also resulted in a 60% drop in problem
reoccurrence and a 50% improvement in root cause analysis and
permanent solution of a problem.
- SDLC
As can be seen, I have had considerable experience with systems development life
cycle (SDLC) disciplines. I have been evolved, in a hands-on
fashion, with each and every phase of the process, from the
enterprise level (see Enterprise Architecture),
the operational level (see Systems Management)
and the development level.
SDLC was initially a concept introduced by the computer aided
software engineering (CASE) tools of the 1980's and early 1990's.
When systems development became an applied science, we needed a
methodology that a computer could model and adhere to. Over the
years SDLC has been systematically applied to everything from
physical infrastructure to internet web sites. "Life
Cycles" are also now present in marketing, business development and
most fields of engineering.
- Summary
My experience does not stop there. I
have done literally hundreds of little 'solutions' for a wide range
of problems. Recent development has turned to web based
solutions and internet issues - like the simple TCP/IP ping and
tracert program that insures a web server stays connected, and if
disconnected seeks out the problem root - local, DNS or ISP... It took 50
lines of code and answers the question instantly.
Development is probably my favorite part
of advanced consulting. It is often thought that a consultant with
as many years in this business as I is more of a conceptual thinker
than a technical practitioner. This is not the case with me.
I would rather do it than talk about it; and rather solve the
problem than the symptom. It's knowing the difference that my
experience affords.
|
