|Project Number: MQP MXC-W076
Process Controls and Improvements
For Deutsche Bank’s Job Execution Framework
A Major Qualifying Report: submitted to the faculty of
WORCESTER POLYTECHNIC INSTITUTE
In partial fulfillment of the
Degree of Bachelor of Science
In cooperation with:
December 15, 2006
Professor Michael Ciaraldi, CS Advisor
Professor Arthur Gerstenfeld, IE Advisor
Professor Jon Abraham, MA Advisor
This report, prepared for Deutsche Bank’s Global Exchange Services (GES) division, is focused on assessing the current job scheduling systems used to execute and deliver reports throughout the bank’s futures and options business. Through a series of interviews, research, and analysis, we developed recommendations to improve the future application, the Job Execution Framework (JEF), and established controls to help regulate the job submission process.
Our group would first like to thank Gregory Friel for making this project possible at Deutsche Bank. Without his continued support and personal dedication, we would not have had the opportunity to work at one of the world’s top investment banks. In addition, we appreciate the use of the bank’s office and services throughout our time here in New York.
We would like to thank our colleagues in London, particularly Ian Ramsay and John Hawkins. Their expertise and excitement toward this project has helped us present effective and strong arguments. In addition, they provided us with access and guidance to valuable technical resources and personal contacts. We appreciate the time they set aside for this project including many hours for conference calls to help us along the way.
We would also like to thank Antonella Allaria for working closely with us throughout this project. Her help coordinating meetings between London and New York made adjusting to the global atmosphere here at Deutsche Bank much easier. Also, the input and background perspective she provided was invaluable to the success of our project.
Special thanks also to Andrea Rudnick, Vilas Hirani, Pete Lindsay, Jens Balkmann, Dennis Klocke, and Iliana Dimitrova for their contributions to our project. Their input through interviews and discussion provided us with important findings to help build the most encompassing and useful recommendations.
Finally, we would like to thank our advisors at WPI, Professors Art Gerstenfeld, Mike Ciaraldi, and Jon Abraham. Their constant support, revisions, and general input to this project helped us to present the best final product. Also, their efforts to coordinate a project center of this magnitude with the world’s top companies have not gone unnoticed, and have provided us with an excellent learning experience.
Within Deutsche Bank’s Global Exchange Services (GES) division, thousands of trades are conducted on a daily basis. Responsible for trading and clearing derivatives within the global Futures and Options markets, GES depends on a set of job scheduling systems to facilitate the flow of data. Currently, several different scheduling systems are used to meet the company’s operations throughout the 73 nations that Deutsche Bank is located. The largest system, TaskMan, is primarily used in the United States and England and to date contains over 2,500 individual jobs, far more than originally conceived. Additionally, two other systems – RANtask and the web-based Asia-Pacific Perl Scripts – are implemented in offices including Frankfort, Germany and Sydney, Australia respectively. However, in an effort to keep up with the ever-growing demand for these systems worldwide and provide Deutsche Bank with one all-encompassing job scheduling system, GES IT developers are producing a replacement system known as the Job Execution Framework (JEF). This new system JEF is being developed as a multi-threaded system with a main goal to allow for future expandability, and is. P planned to officially replace the current systems by late 2007, early 2008. Prior to its deployment however,, developers must identify and address all the current problems, and adjust JEF accordingly. This project focused on identifying and analyzing the current issues in the job scheduling systems at the bank, notably the large TaskMan system, to present recommendations to improve and safeguard JEF against similar problems.
In order to accomplish this goal and provide Deutsche Bank with the most effective recommendations, this project achieved the following four main objectives:
Identified Job Submission process and problem areas
Established controls and new submission process to apply to JEF
Determined job migration steps and controls to identify active jobs to transfer to JEF
Recommended features to implement in JEF
To accomplish these tasks, we first conducted research on the current job scheduling systems (primarily TaskMan) through a series of interviews with employees in GES IT, Operations, and Management. By utilizing the different perspectives and expertise of each person within these divisions, it was possible to piece together the complete job submission process. With a step-by-step process documented, we easily identified problem areas and potential improvements that could be made to the Job Execution Framework. In particular, we constructed a comprehensive list of limitations and issues that plague the current applications used, including TaskMan, the Global Incident Management System (GIMS), RANbase, and RANtask. These issues – the basis of the majority of our recommendations – were prioritized in order to allow GES IT to easily address the problems with the greatest impact. Examples of TaskMan problems include:
A lack of controls
No direct link between GIMS ticket and TaskMan
Users cannot view or search what jobs are already running
No standardized testing system established prior to job submission
No method for users to track the status of a job to see if it is working properly
Many of these issues are attributed to the fact that TaskMan was originally constructed to simply run eight to ten Structured Query Language Scripts (referred to as jobs/queries). However, in order to run the 2,500 jobs now in the system, TaskMan was modified in an ad-hoc manner with minimal regard to future issues and expandability. Therefore, GES IT must understand the benefits, limitations and functionality of the current job scheduling systems to improve the features of the Job Execution Framework.
As mentioned, a lack of controls is the most evident problem in the job submission process. There is no governance system to regulate who is qualified to submit a job into the system, which allows poorly written, unnecessary, and often duplicate jobs to enter. In addition, few controls are established to provide quality testing for each new job. The role of Operational Excellence (OE), the group within GES IT that currently inputs jobs into TaskMan, has been reduced to nearly a middleman. Rather than formally testing the performance of jobs, jobs are entered on an inconsistent case-by-case review process.
After analyzing the results gathered, we were able to form many recommendations to prevent problems from occurring as Deutsche Bank transitions to the JEF system. We focused on providing recommendations on the following areas:
Job Submission Requirements
Role of Operations and Information Technology
Business Objects as an Alternative
Job Migration to the Job Execution Framework (JEF)
Standardized Reporting Application
Searchable Job Database
Job Execution Frameworks Improvements
These recommendations are prioritized to provide Deutsche Bank with an understanding of their potential improvements.
In order to efficiently submit and track each job that enters JEF, we recommended that each new job submission or modification contains basic standardized information. Also, all data gathered should be stored in the expandable XML database used by JEF to ensure that proper contact information and job descriptions are readily accessible in the event of future problems or maintenance. During this process, we recommended that users do not assume the responsibility of including their own SQL queries with job submission. Rather, in order to clearly designate the roles of Operations and GES IT, users should build “business requirements” for IT to interpret into the most effective SQL queries. This restructuring, along with a searchable database of jobs, will minimize the number of repeated and poorly running jobs that enter JEF. In addition, Deutsche Bank should also provide different levels of access rights for the new job scheduling system. We recommended that access levels are established including:
Update and write access to input/change configurations (reserved for IT)
Read-only access to view status of job (available to IT and qualified users)
In TaskMan, virtually anyone who wishes to submit a new job request can fill out a Global Incident Management ticket – regardless of their position, experience, or need for the job.
Next, to provide an appropriate method to migrate only the active, needed jobs from TaskMan and other systems to JEF, we recommended that GES IT study each job on a case-by-case basis. Only once adequate contact information and a job description are available, should a job be transferred to JEF. This report provides guidelines for the specific procedures that GES IT should follow during this migration. Also, to ensure that all of the required information is provided with each new request in JEF, we proposedrecommendsuggested developing a standardized reporting tool, a proposed wizard, and reverting GIMS back to its original purpose as a support mechanism. Currently, GIMS allows for too much individual customization during the job submission process to effectively capture the recommended data. The wizard is also designed to help create a separate searchable library of jobs within JEF. All of the information gathered during submission, including a detailed job title and purpose, should be stored in a database to allow a user to search for existing jobs based on specified parameters. Finally, to improve the capabilities of the support team, we recommended creating a more robust logging system and more effective alert system in the JEF system. These recommendations were designed to smoothly transition to the Job Execution Framework and will help to ensure the continued success of the new job scheduling system into the future.