Managing Application Software Maintenance

Presented at the Masdar Institute in
Abu Dhabi on May 1, 2012 by
Dr.Parasuram Balasubramanian,
C.E.O.,Theme Work Analytics,
India and USA
Application Software Maintenance- A
Queuing Theory approach for
continuous improvement
balasubp@gmail.com

Application Software Maintenance
• 70 % of the I.T.software budget in any large
company is spent on Application Maintenance
• They are ERP, Financial Accounting, Costing,
SCM ,CRM , Compliance Reporting or Share
holder servicing systems .
• They could be either custom developed ,
bought out packages or hybrid
• Even when packages are implemented, they
tend to be customized over a period of time
• An average system will have close to a million
lines of COBOL code or equivalent
• These systems need to be maintained on an
ongoing basis.

• Maintenance activity can be tedious since the
system documentation lags its development ; does
not contribute to technology skills augmentation
for staff
• Such applications also carry the business risk of
interruption when key staff leave the company
• Yet they are mission critical systems. They need
to be maintained.
• Firms in India, Ireland and Philippines have found
a service market here and have grown rapidly

The Maintenance need arises due to
• Bugs undiscovered in initial implementation
• Lax data verification procedures and subsystems
• Changed data conditions
• Changing business needs
• Changes in environment; legislation etc
• Bugs injected during the Maintenance process

The Maintenance effort required is a function of
• Bugs density
• Total lines of code
• Frequency of use of the system etc
• Required service level
Hence a critical challenge facing firms is to manage the code integrity and the
resources expended to maintain the system. This has to be done in the
context of the task being mundane with low levels of worker motivation.

The Maintenance initiatives are categorized as
1. Minor bugs
2. Major bugs ( minor or major determined by
function criticality and effort required to rectify)
3. Enhancement initiatives ( minor or major)
The maintenance team size is usually constant. Hence the challenge facing
the team manager is to decide on allocation of resources to different
maintenance tasks. When an enhancement initiative is time critical and
requires large effort level , the work is either outsourced or team size is
increased. Most of the time the Maintenance team works with a back log.

In earlier years we were faced with the task of
submitting a competitive proposal to Nynex
corporation to take over the maintenance of
number of key applications.
Queuing Theory based approach was evolved at that time.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Ideally, the service capacity should be higher than the service demand
to avoid build up of queues and extensive wait time in queues.
It should facilitate first in first out principle.
In reality they are not assured.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Servers tend to specialize in parts of the application.
Arrival rate at times can be excessive. Demand> capacity
Randomness of arrivals leads to queue build up
Item rectification priority changes dynamically.
Bins

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
We determined that key to our success in winning the bids is
to improve server productivity.
The productivity improvements should result in decrease in
number of servers needed from time to time.
Bins

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
We separated the efforts needed for bug fixing from enhancements.
We could reduce the team size from quarter to quarter if
(a) The customer agreed to freeze the enhancement requests for a
period of time or to treat them as a separate project
(b) We can improve server productivity continuously
Bins

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Server productivity can be improved by
• Building multiple skills in each server
• Bringing the documentation upto-date
• Creating a shared knowledge data base for system based learning
• Creating standard procedures for error analysis, impact analysis, rectification
and testing
Bins

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Server productivity can be improved by
• Use of domain experts as trainers and project resources
• Extensive use of metrics to measure and report productivity levels and
service levels
• Restructuring the code periodically to improve its maintainability
Bins

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
When this approach was articulated in our proposal and backed by commitment to
reduce resources needed for maintenance from quarter to quarter
• Nynex was impressed sufficiently to hand over 3 major systems for a period of
three years.
• Successful execution of these projects led to our winning many assignments from
other customers with similar needs. The project data were freely shared with
customers .
Bins

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Post Y2K the pressure on use of IT dollars for development projects was
high. Customers became very demanding in aggressive reduction of
resources used for maintenance and redeployment in development
projects .
Bins
This challenge was converted into an opportunity by the following
approach.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
A causal analysis of the arrival rate was carried out of previously executed
projects. It revealed that (a) slow surfacing of hidden bugs over a period
of time and (b) bugs injected during the maintenance process were the
prime causes.
Bins
We sensed that a dramatic solution would be to keep reducing the arrival
rate continuously.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Till then testing was confined to
• Successful removal of the reported bug
Bins
It called for a radically different approach to testing the software.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
In future , we proposed
• Regression testing to ensure that no other part of the system was affected
due to this change
• Comprehensive testing of the system in the first quarter ( as though customer
was contractually obligated to deliver an error free system to us )
Bins
It called for a radically different approach to testing the software.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Bins
Further the data edit programs and modules were to be reviewed and
tightened to avoid incorrect data getting through far into the system and then
causing system halts.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Bins
These led to early capture of hidden bugs and minimizing of freshly
injected bugs. The benefits realized were substantial. Maintenance team
size , in some instances went down from 12 to 2 over a two year period.
The happy customer was more than willing to give the development
assignment to the team members released from maintenance duty.

Multiple
servers
Arrival of
maintenance
requests
Rectified/
enhanced system
Bins
Thank You.
balasubp@gmail.com

Managing Application Software Maintenance

More Related Content

What's hot

Viewers also liked

Similar to Managing Application Software Maintenance

More from Parasuram Balasubramanian

Recently uploaded

Managing Application Software Maintenance