Case study-loan-processing-test-plan

Loan Processing Test Plan | RBCS
Overview
The “Some Loan App”, as it will be deployed in Release 1.0, allows Home Equity Loan
Call Center Agents to fit home equity products (loans and lines of credit) to customers.
At a high level, the system is configured as shown in figure one below. The “Some Loan
App” itself is a group of Java programs and assorted “glue” that run on the WebLogic
server. The Oracle database server provides storage as the application is processed, while
the Netscape server offloads gateway activities to the clients from the WebLogic server.
From a systems perspective, the “Some Loan App” can be said to “work” if it enables
Call Center agents to initiate, negotiate, and close loan transactions with desirable
customers while helping these agents avoid issuing loans to people unlike to pay them
back.
Figure 1: “Some Loan App”, Release 1.0
Loan transactions “flow” through the “Some Loan App” in the following fashion. Call
Center agents in the Fairbanks Call Center receive phone calls from potential customers.
They interview the customer, entering information into the “Some Loan App” through a
Web browser interface. At a certain point, the “Some Loan App”, talking to the Scoring
mainframe through MQ services, scores the customer’s credit. Based on this credit score,
the “Some Loan App” displays various products that the Call Center agent can offer to
www.rexblackconsulting.com Copyright (c) 1998-2002 Rex Black All Rights Reserved April 10, 1998

Loan Processing Test Plan | RBCS DRAFT Revision 0.9 Page 2 of 26
the customer. If the customer chooses one of these products, the Call Center agent will
conditionally offer the loan. The interview ends and “Some Loan App” transmits the loan
information to “Some Other Loan App” for origination. Ultimately, though the services
of the back-end server and documentation generator, the customer receives her loan
documents.
The following test plan describes the testing to be performed by the Minneapolis Test
Group (MTG), Somebank’s independent test team for “Some Loan App” testing. This
test plan covers the included items in the test project, the specific risks to product quality
we intend to address, timeframes, the test environment, problems that could threaten the
success of testing, test tools and harnesses we will need to develop, and the test execution
process. Some testing occurs outside of the independent test team’s area, such as unit
testing. In general, our testing includes positive use cases developed by Robert S.,
Assistant Vice President, Home Equity Group, and capacity and performance test cases.
Bounds
The following sections serve to frame the role of the independent test organization on this
project.
Scope
The following table defines the scope of the MTG independent test effort.
MTG Independent Test (Release 1.0)
IS IS NOT
Positive use cases (functionality)
Capacity and volume
Error handling and recovery
Standards and regulatory compliance (as
covered in the use cases)
Client configuration (browser and call
center desktop compatibility)
Security [TBD by Todd R. 7/22]
Distributed (leverage Webdotbank testing)
Performance
Black-box/behavioral testing
“Some Loan App”/”Some Other Loan
App” status communications
Confirmation testing in QA region
Negative use cases
Operations (e.g., paperwork processing,
loan initiation, rate updates, etc.)
Usability or user interface
Date and time processing
Localization
Test database development
Documentation
Code coverage
Software reliability
Testing of the complete system
Horizontal (end-to-end) integration
Data flow or data quality
Unit or FVT testing
White-box/structural testing
Table 1: MTG independent test team IS/IS NOT (scope)
Clearly, the test team could do more, but we do not want to raise the quality bar for the
“Some Loan App” for Release 1.0 at this stage in the game. Subsequent test efforts—i.e.,
Release 1.1 testing—will include many of the items in the “IS NOT” column above.
Definitions
The following table defines commonly used test terms and other terms found in this
document.
Term Meaning
Acceptance Test A set of tests, usually selected from the System Test suite,

Term Meaning
chosen to demonstrate that the System Under Test (see
below) meets reasonable quality standards.
Black Box Testing Testing based on the purposes a program serves; i.e., behavioral
testing.
Bug Some aspect of the system under test that causes it to fail to meet
reasonable expectations. “Reasonable” is defined by iterative
consensus if it is not obvious.
Build A collection of software objects of known revision levels
compiled into one or more software executables for
installation on the system under test.
Code Promotion Installing a software build, in the form of a JAR file, into a
server region. Also known as “build apply.”
Confirmation Test A selected set of tests designed to find ways in which a bug fix
failed to address the reported problem fully.
Entry (Exit)
Criteria
The parameters that determine whether one is ready to enter
(exit) a test effort.
Integration Test In this plan, a set of tests designed to find bugs in typical
horizontal paths through integrated system components.
JAR File Java archive containing a Java program.
Oracle A method or procedure, often integrated into a test tool, for
determining whether the system under test is behaving
correctly. This can involve examining all the outputs or a
sample of the outputs, either from a user perspective
(behaviorally) or from a system internals perspective
(structurally).
Quality Risk The possibility of a specific system failure mode, either
localized, caused by subsystem interactions, or a knock-on
effect of a remote system failure, that adversely affects the
system’s user.
Reference Platform A “known correct” system against which one can compare the
results and behavior of the system under test. In this specific
case, “Some Other Loan App” is the reference platform in
terms of products offered, while the “Some Loan App” is the
system under test.
Region A set of servers configured to perform a particular role in terms
of development, testing, or production.
Regression Test A selected set of tests designed to find new failures, or
regression, that changes, usually associated with bug fixes,
have caused in subsystem, interface or product functionality.
Smoke Test A limited set of regression tests (see above) designed to
determine, though a random sample of critical functions,
whether a given build is ready for testing. For this project,
Smoke Test also includes confirmation tests (see above) in
the Integration Region.
System Test A set of tests designed to find bugs in the overall operation of the
integrated system.

Term Meaning
SUT System under test. In this case, the Call Center agent desktop and
browser, the HTML transmitted to the browser, the Netscape,
the WebLogic, and the Oracle servers, the Scoring and
“Some Other Loan App” interface modules, and the network
infrastructure.
Test Escape A field failure that could reasonably have been found by MTG
executing this test plan but for errors in execution,
attentiveness, interpretation of observed behavior, or other
such problems.
Validation Test A complete set of regression tests, including final confirmation
of bug closure.
White-Box Testing Testing based on the way a program performs its tasks; i.e.,
structural testing.
Table 2: Definitions
Setting
The test efforts described in this plan will take place in the following locations. (See
Human Resources below for a description of the people in this table.)
Location Test Effort Staff
“Some Client”
(Minneapolis)
System test execution. Minneapolis Test Group
RBCS
(Bulverde)
Test project management.
Test team staffing
Rex B. (off-site two days
per week)
Webdotbank
(Seattle)
Testing of the software
components provided.
Webdotbank Test Team
Table 3: Locations involved in testing
Quality Risks
The test cases that will be used by the test team were primarily developed by the business
in Fairbanks. As such, MTG doesn’t have a lot of visibility into the quality risk
management provided by them. MTG did develop test cases to cover specific bugs found
during the Beta I process in Fairbanks; however, these do not correspond to specific
quality risks, but rather to failure modes observed.
Schedule of Milestones
The following shows the scheduled events that affect this test effort.
Milestone/Effort Start End
Unit test complete 4/22/98 4/22/98
Smoke Build delivered and installed 4/23/98 -
System Test Entry Criteria Met
4/24/98
-
System Test (six release cycles) 4/27/98 6/5/98
System Test Launch Meeting 4/27/98 -
Build 1 delivered and installed 4/27/98 -
Build 3delivered and installed 5/11/98 -

Milestone/Effort Start End
Golden Code review (all bugs fixed: ready for final build) 5/29/98 -
System Test Exit Criteria Met 6/5/98 -
System Test Phase Exit Meeting 6/5/98 -
User Acceptance Test 6/8/98 6/19/98
Go/No-Go Decision 6/22/98 -
Deployment 6/23/98 7/2/98
Table 4: Scheduled System Test milestones
Transitions
The following subsections define the factors by which project management will decide
whether we are ready to start, continue, and declare complete the test effort.
System Test Entry Criteria
System Test can begin when the following criteria are met:
1. The “Tracker” bug tracking system is in place and available for all project
participants.
2. All software objects are under formal, automated source code and configuration
management control.
3. The HEG System Support team has configured the System Test clients and servers
for testing, including the “cloned” call center agent desktops, the LoadRunner Virtual
User hosts, the Netscape server, the WebLogic server, the Oracle server, the Oracle
database tables (including indices and referential integrity constraints), the MQ
connections, and the network infrastructure. The Test Team has been provided with
appropriate access to these systems.
4. The Development Teams have code-completed all features and bug fixes scheduled
for Release 1.0.
5. The Development Teams have unit-tested all features and bug fixes scheduled for
Release 1.0, and transitioned the appropriate bug reports into a “verify” state.
6. Less than ten (10) must-fix bugs (as determined by Magdy J. , “Some Loan App”
Release 1.0 Project Manager, and Robert S., Assistant Vice President, Home Equity
Group) are open, including bugs found during unit testing.
7. The Development Teams provide revision-controlled, complete software products to
MTG. (See the “Release Management” section.)
System Test Continuation Criteria
System Test will continue provide the following criteria are met:
1. All software released to the Test Team is accompanied by Release Notes. These
Release Notes must specify the bug reports the Development Teams believe are
resolved in each software release.

2. No change is made to the “Some Loan App”, whether in source code, configuration
files, or other setup instructions or processes, without an accompanying bug report.
3. Twice-weekly bug review meetings occur until System Test Phase Exit to manage the
open bug backlog and bug closure times.
System Test Exit Criteria
System Test will end when following criteria are met:
1. No panic, crash, halt, wedge, unexpected process termination, or other stoppage of
processing has occurred on any server software or hardware for the previous three (3)
weeks.
2. The Test Team has executed all the planned tests against the GA-candidate software
release.
3. The Development Teams have resolved all must-fix bugs. Magdy J., “Some Loan
App” Release 1.0 Project Manager, and Robert S., Assistant Vice President, Home
Equity Group will define which bugs are must-fix.
4. The Test Team has checked that all issues in the bug tracking system are either closed
or deferred, and, where appropriate, verified by regression and confirmation testing.
5. The open/close curve indicates that we have achieved product stability and reliability.
6. The Project Management Team agrees that the product, as defined during the final
cycle of System Test, will satisfy the Call Center Agent’s reasonable expectations of
quality.
7. The Project Management Team holds a System Test Phase Exit Meeting and agrees
that these System Test exit criteria are met.
Test Configurations and Environments
Testing involves both client systems and server configurations. There are two types of
client systems present in the lab.
• Load Runner Virtual User clients (“LR clients”). These run Windows NT, have 128
MB of memory, and use Pentium 300 MHz CPUs. [Clarence: Please confirm
configuration.] These systems are configured to support large numbers of
simultaneous Load Runner “Virtual User” sessions.
• Call Center Desktop Agent clients (“CC clients”). These run Windows 95, have 64
MB of memory, and use Pentium 200 or better MHz CPUs. [Clarence: Please confirm
configuration.] These systems are configured to resemble, as closely as possible in the
“Some Client” network environment, the “Some Client Partner” Fairbanks Call
Center Agent Desktop configuration.
The LR clients are used to run stress, performance, and capacity test cases, while
the CC clients run the manual test cases. We have two LR clients in the test lab,
two other “spare” LR clients under Clarence T.’s control, and six CC clients in the
test lab.
These systems can connect to a variety of server configurations known as
“regions.” The following three regions are used during System Test.
• “Some Loan App” QA Region. This is the server environment into which the CC and
LR clients send the “Some Loan App” loan applications during testing.

• Scoring QA Region. This server environment provides credit bureau scoring to the
“Some Loan App” so it can decide which tier the customer suits the customer’s credit
risk.
• “Some Other Loan App” Regression Region. Once the tester has entered a test
application into the “Some Loan App” in the QA region, the “Some Loan App” sends
the loan to this “Some Other Loan App” server environment. After the “Some Loan
App” sends the loan here, the tester will verify that the loan is present in the system
and was correctly received by logging into “Some Other Loan App”. He will then
compare the correctness of the products offered by the “Some Loan App”, by entering
the same application into this “Some Other Loan App” server environment, then
comparing the offered products.
Figure 2 below shows the various regions, how the tester interacts with them, and
how they interact with each other.
The various regions used by the Development Team and MTG are accessed through the
client system’s Web browser. The browser is pointed to the correct region via a URL
(favored approach) or an IP address (disfavored approach). The following table describes
these access methods.
Region URL IP
“Some Loan App”
Integration
http://hegtest.bank.c
om
168.118.19.205
“Some Loan App”
QA
http://hegqa.bank.co
m
168.118.19.221
168.118.19.222
Table 5: Regions used by Development and QA

Figure 2: Regions used by testing
As discussed elsewhere the Test Plan, the “Some Other Loan App” Regression Region
serves as the reference platform. “Some Loan App” results, in terms of products offered,
should match the results from “Some Other Loan App”. Because of the handling of
duplicate loan applications, “Some Other Loan App” will re-use the credit information
provided to it by the “Some Loan App”, meaning that the reference platform results are
not entirely independent of the System Under Test results. It is conceivable that the
“Some Loan App” could corrupt the scoring information in such as way as to cause
“Some Other Loan App” to offer the same (wrong) products. If we have a test escape
because of this risk, this problem will be caught in UAT when Call Center agents enter
“live” applications into “Some Other Loan App”. Finally, it is not possible to use “Some
Other Loan App” as a reference platform in terms of performance, error handling, and so
forth, because the behaviors of Unix character-based and Web GUI-based applications
are so different.
Test Development
The bulk of the manual testing performed for Release 1.0 consists of UAT test scripts
developed by Robert S. and the Enormii Consulting Firm team as part of the original
UAT/Integration Test effort. No new development is required, other than changes to
these existing scripts to accommodate bug fixes required for defects found during the
Beta One exercise in December.
Some of these bug fixes, however, required entire new areas of functionality. (These bugs
were actually design changes rather than mere defects in the existing function.) We have
implemented a number of new manual test cases to cover these changes. These test cases
have been provided to Robert S. for review and approval.

Robert S. is also considering developing some additional tests to cover certain areas left
untested by the original test scripts. If she does so, we will request these scripts in order
to add them to our manual test suite.
Finally, Emma G. is working with Julius R. to redefine the automated Load Runner
scripts that carry out the stress, capacity, and performance test cases. These efforts are, at
this time, incremental; wholesale re-engineering of the automated suites will be part of
the Release 1.1 effort. The most significant changes to the suites in terms of the tests
performed arise from using more realistic time-per-page wait periods. These more-
accurately model the time Call Center agents will spend filling in each page prior to
submitting it to the “Some Loan App” servers.
Test Execution
The following subsections define the activities and participants involved in test
execution.
Test Hours
MTG will test primarily between the hours of 9:00 AM and 5:00 PM (all times EST),
Monday through Friday. However, we may start earlier and continue later, as the
workload requires. The escalation process section below specifies on-hours and off-hours
support numbers.
Test Cycles
A test cycle will begin with each weekly release of a new software build as discussed
elsewhere in this plan. Test execution will begin as soon as the new releases are
appropriately installed. Testing will be carried out in the following order:
• Confirmation testing. The test team will retest all bugs reported as fixed in the
release notes. For those bugs fixed, the testers change the state of the appropriate bug
reports to “Resolved”. They mark as “Not Resolved” reports for any bugs not fixed.1
• Scheduled testing. The test team will run all test cases scheduled for that test cycle.
Should the test team complete all scheduled test cases, it will perform any other test cases
not scheduled for that cycle but which have not been run against the current build lab. If
those cases are then completed, the test team will perform exploratory (ad hoc) testing
until the following release begins the next cycle. Conversely, due to delays of
deliverables, schedule pressures, high bug find rates, or the size (duration or effort) of the
test suites, it may not be possible to complete all the scheduled test cases during every
cycle. In that case, the test cases not completed will be rescheduled as first priority for the
next cycle, immediately following the confirmation testing.
Test Execution Process
The objective of the System Test process is to find bugs in the system under test. We
intend to do this by running a set of manual and automated test cases against each build.
1 In addition to testing performed by MTG in the QA Region, the Development Team will
perform a daily confirmation test of the nightly builds provided by Webdotbank in the Integration
Region. For the tests that fail, they will re-open the appropriate bug reports and return them for
further work. MTG will not perform confirmation testing of these bug fixes since they have
already failed in the Integration Region. For the Development Team tests that pass, they will re-
assign the appropriate bug reports to Jenna B., who will coordinate the MTG confirmation testing
in the QA Region.

We want to run every test case at least one time, and ideally four times, as part of the
System Test Phase. To do so, we will use the Test Case Summary worksheets discussed
below. At the beginning of each test cycle, we will assign a “basket” of test cases to each
tester. Each tester’s basket will be different from one test cycle to the next, to ensure that
any invalid assumptions on one tester’s part do not result in a test escape. Once assigned
their basket of tests, these testers will follow the process described in the rest of this
section to execute each test case, and repeat the process until they have exhausted their
list. If they empty their basket prior to the end of a cycle, they are to assist other testers
by coordinating a reassignment of some of those testers’ test cases to themselves. If all
tests are completed, the test team will proceed as described in the section above.
Because testers must switch back and forth from the “Some Loan App” to “Some Other
Loan App” to confirm correct product offerings, testers will run five scripts at once. The
process is as follows:
• Log into the “Some Loan App” using personal ID, if it’s the first batch of test cases
for the day.
• Print five of the test scripts from the assigned basket.
• Enter the test scripts into the “Some Loan App” one at a time, recording the Fit Loan
ID numbers as they appear.
• Record if the application is approved or declined. If one or more application is
declined, and such are the expected results, the testers need not enter these
applications into “Some Other Loan App”.
• Log on to “Some Other Loan App” and find the loans transmitted by the “Some Loan
App”. Verify that they were transmitted correctly.
• Re-enter the identical test scripts into “Some Other Loan App”.
• Record the “Some Other Loan App” APP number.
• Verify that the Credit Bureau response in “Some Other Loan App” matches the
“Some Loan App”’s response, specifically if the loan is approved or declined.
• Verify the products offered in “Some Other Loan App” match the “Some Loan App”
offerings, especially in terms of interest rate, terms and conditions, and amount.
• Prepare one or more bug reports immediately upon locating any discrepancies
following the process described later in this Test Plan.
• Report status of the five test cases to Jenna B., Test Coordinator/Engineer, so that he
can update the Test Case Summary spreadsheet for that cycle.
• Repeat this process for the next set of five test cases, or log out of the “Some Loan
App” if this set was the last set for the day.
To ensure proper testing, the following parameters apply to this process:
• While testers will share “Some Other Loan App” login IDs, they will not “share”
“Some Loan App” login IDs.
• Testers are expected to log out from the “Some Loan App” at the end of each day or
before going on a break longer than half an hour.
• If a tester comes across an unattended CC client in the test lab logged into the “Some
Loan App”, the tester should log that client out of the “Some Loan App”, especially if
he or she plans to use that CC client to perform a test.

• Testers are to use only CC client systems for manual testing. Testers are not to use
their personal desktops for manual testing, as they are not configured the same way as
the Fairbanks Call Center Agent Desktops. LR clients are reserved for automated
testing via Load Runner.
Automated testing proceeds as follows. Emma G., in consultation with Julius R., will
finalize a set of performance and capacity test scripts. (See below for stepwise plan.)
Clarence T. will then submit the LoadRunner jobs. (Clarence must be the “trigger man”
because of “Some Client” IS Department concerns about errant LoadRunner processes
flooding the production T1 that connects the Minneapolis office with the rest of “Some
Client Partner”.) Once the test runs are complete, Emma and Julius will analyze the
results.
Based on previous findings, we do not expect the “Some Loan App” to satisfy the
performance and capacity goals on the first run. Therefore, we will pursue a stepwise
approach. We plan on five “steps.” We hope to take one or more steps in each weekly
cycle. Each step is a single type of test case. There are four other categories of test cases,
based on accepted or declined applications and joint or single applications. Therefore, we
have twenty performance test cases, as shown in the table below.
Single Joint
Simultaneous
Users
Application
Accepted
Application
Declined
Application
Accepted
Application
Declined
20 Users Test Case 1 Test Case 2 Test Case 3 Test Case 4
Table 6: Stepwise approach to capacity and performance tests
Because this testing, like the rest of the testing, must take place in the QA Region to be
meaningful—this is the region that actually resembles the Production Region—
Development cannot use this environment for debugging Performance problems by
dropping in nightly builds once System Test proper begins. We recommend that the
Development team use mathematical models to ensure that its performance-affecting
design changes will enable sufficient performance before “burning” one of the four
windows in which to run performance tests.
Human Resources
The table describes the human resources need to execute this plan. Key support and
liaison roles are also defined.
Title Roles Name
Test Manager Plan, track, and report on test execution
Secure appropriate human and other
resources
Provide technical and managerial
leadership of test team
Audit test processes and results of
external test participants
Rex B.
Test Develop manual tests Jenna B.

Title Roles Name
Coordinator
/ Engineer
(Manual)
Execute manual tests
Train the other test engineers on manual
tests
Track the test cases as discussed
elsewhere in this plan
Test
Engineers
(Manual)
Execute manual tests
Report results to the Test Coordinator
Hemamalini D.
Cheri L.
Greg J.
Test System
Administrator
Ensure proper configuration of all the test
lab workstations.
[TBH]
Test Toolsmith Design and implement LoadRunner test
suites
Interpret LoadRunner test results
Emma G.
HEG Level
One Support
Provide first-line problem-escalation and
test system configuration support to
MTG for all regions involved in the
System Test Effort
Escalate problems to “Some Loan App”,
Scoring, “Some Other Loan App”, and
SRM support as necessary (see
process below).
Clarence T.
HEG Level One
Support Manager
Manage the support and escalation
processes
Mac D.
“Some Loan App”
Manager
Coordinate with the Test Manager the
“promotion” of weekly “Some Loan
App” releases into the “Some Loan
App” QA Region.
Hanna R.
“SOME OTHER
LOAN APP”
Release Manager
Coordinate with the “Some Loan App”
Release Manager the “promotion” of
as-needed “Some Other Loan App”
changes into the “Some Other Loan
App” Regression Region.
Sam C.
Scoring Release
Manager
Coordinate with the “Some Loan App”
Release Manager the “promotion” of
as-needed Scoring changes into the
Scoring QA Region.
Jenny V.
SRM Support Deal with infrastructure issues in the QA
Region.
Petra V.
SRM Build
Manager
Promote the correct build weekly into the
QA and Production Regions.
Petra V.
Table 7: Human resources for testing
Escalation Process
Unexpected failures, logistical challenges, and strange behavior occur during System Test
efforts. Some of these events can block forward progress through the test suite. The

following is the escalation plan when such blocking events occur during normal Test
Hours. [Mac D./Magdy J.: Is this closed now?]
1. MTG shall escalate to Clarence T. any blocking issues encountered.
2. Clarence T., working with the appropriate Webdotbank, “Some Other Loan App”,
Scoring, and other resources, shall ascertain if the problem is an infrastructure issue.
If so, he shall resolve the problem. (If not, see step four.) Status updates shall go out
via e-mail at least every hour. At least the following people shall be on the
distribution list, and, where these people have multiple e-mail addresses listed below,
messages shall be sent to both addresses.
• Magdy J.
• Rex B.
• Emma G.
• Harry W.
• Jenna B.
• Hanna R.
3. Once Clarence has resolved the problem, he shall notify at least one member of the
MTG via the phone numbers shown below that the problem is resolved, as well as
sending out a final e-mail status update. (Steps four through seven are unnecessary if
Clarence resolves the problem.)
4. If, after evaluation, Clarence concludes the infrastructure is functioning properly,
thereby implicating the “Some Loan App” and its associated components as the root
problem, Clarence shall hand off the problem to Jenny V. for resolution. Clarence
shall pass the issue, his findings and the MTG contact person’s name to Jenny V. He
shall also inform participants of his conclusions and that Jenny V. has taken over the
issue as follows:
• The MTG contact person via the phone numbers shown below; and,
• The e-mail distribution list shown above via e-mail.
5. Jenny V., working with the appropriate Webdotbank, “SOME OTHER LOAN APP”,
Scoring, and other resources, shall resolve the problem in Fit, “SOME OTHER
LOAN APP”, Scoring, or wherever it resides. Status updates shall go out via e-mail at
least every hour. At least the following people shall be on the distribution list.
• Magdy J.
• Rex B.
• Emma G.
• Harry W.
• Jenna B.
• Hanna R.
6. Once Jenny V. has resolved the problem, he shall notify at least one member of the
MTG via the phone numbers shown below that the problem is resolved, as well as
sending out a final e-mail status update.
7. If Jenny V. feels it is appropriate, he shall instruct Clarence to rollback the code so
that testing can continue. (This should only be done as a last resort.) Clarence shall

inform Jenny V. upon completion of the rollback and Jenny V. shall close the issue as
described in step six above.
During off-hours, the following process shall apply:
1. MTG shall escalate to the SRM duty phone any blocking issues encountered. The
number left on the pager is the SRM contact number, so the MTG team member shall
leave a cell phone number or a desk phone number with voice mail. (The MTG
member shall only leave a voice mail phone number at which he or she can
personally retrieve messages.)
2. Once MTG receives a return call from SRM, the MTG team member escalating the
issue shall describe the problem, then exchange contact information for future follow
up.
3. SRM shall provide regular updates on status, then an all-clear once the problem is
resolved. If more than thirty minutes goes by without status, the MTG team member
escalating the problem shall contact Rex B. via mobile phone. Rex B. may dismiss
the staff for the remainder of the off-hours period—which may be the entire weekend
—if SRM does not respond in a timely fashion.
Should any person on this list have difficulty, for whatever reason, executing the
appropriate process, or should the process prove incapable of unblocking the problem
within four hours, the MTG staff shall escalate the situation via mobile phone to Rex B.
and Magdy J. Please note that all people named in the process below shall carry a mobile
phone or pager, turned on with adequately charged batteries and sufficient coverage
range, at all times during System Test. Automated and manual testing may—and
probably will—occur continuously during System Test. Periods of unavailability shall be
communicated to Magdy J. , “Some Client” “Some Loan App” Release 1.0 Project
Manager, at least one week in advance, and the person making himself or herself
unavailable shall arrange for alternate coverage.
Test Contact List
Name Role E-mail Office Cell/Pager
Hemamalini
D.
Manual test engineer hd@bank.com (768) 555-
1450 (lab)
N/A
Cheri L. Manual test engineer cl@bank.com (768) 555-
1450 (lab)
N/A
Greg J. Manual test engineer gj@bank.com (768) 555-
1450 (lab)
N/A
Jenna B. Test coordinator/lead
test engineer
jb@bank.com (768) 555-
1450 (lab)
(892) 555-1890
(c)
(892) 555-0945
(p)
Emma G. Test Toolsmith eg@bank.com N/A (434) 555-6789
(c)
Rex B. Test manager Rex_Black
@acm.org
(830) 438-
4830
(210) 241-8209

Support Contact List
Clarence T. System administrator ct@bank.com (325) 555-
1790
(888) 555–1234 (c)
(888) 555–9090 (p)
Jenny V. System programmer jv@bank.com (267) 555-
8903
(800) 555-6892
(p)
SRM Hotline SRM support N/A (142) 555-
7894
N/A
Management Contact List
Magdy J. “Some Loan App” R 1
Project Manager
mj@bank.com (460) 555-
8358
(245) 555-9874
(c)
(888) 555-9780
(p)
Mac D. Support manager md@bank.com (465) 555-
9834
(465) 555-2308
(c)
(888) 555-9024
(p)
Hanna R. Webdotbank vendor
manager
hr@bank.com (372) 555-
8010
(215) 555-0945
(c)
Harry W. Project manager hw@bank.com (456) 555-
1345
(222) 555-9085
(c)
Test Case and Bug Tracking
Test cases will be tracked using a set of Excel worksheets. These will provide both detail
level and summary level information on test cases for each test pass. As discussed
elsewhere, we intend to complete four passes through all the tests, so the Excel
spreadsheet file will have eight worksheets, four pairs of Test Case Summary and Test
Suite Summary worksheets. Jenna B., Test Coordinator/Engineer, will maintain and
update these tracking documents. For each test case, we will track the information shown
in the following table.
Column Heading Meaning
State The state of the test case. The possible states are:
Pass: The test case concluded successfully.
Warn: The test case concluded with an error, which the
project management team has either deferred, closed as
external to the product, or closed as unavoidable.
Fail: The test case revealed a defect that development
will address.
Closed: The test case previously revealed a failure that is
now resolved.
In Queue: The test remains to be executed (indicated by
a blank in the column).
Skip: The test will be skipped (explanation required in

Column Heading Meaning
“Comment” column).
Blocked: The test cannot be run (explanation required in
“Comment” column).
System Configurations In most cases, the Workstation ID (from the front of the
case) for the CC client where the tester entered the test
case application, and the “Some Loan App” Region
(usually QA) where the application was processed.
Bug ID If the test failed, the identifier(s) assigned to the bug by
Tracker when the tester entered the report. Jenna B. will
enter each bug ID on a separate row below each test case
row in the Test Case Summary worksheet.
Bug RPN The risk priority number (severity times priority) of the
bug(s), if applicable, in a column next to each bug ID.
Bug State The state (see below) of the bug report, if any, filed
against this test case.
Plan Hours The planned effort (person-hours) for the first execution
of this test case.
Actual Hours The actual duration (person-hours) required.
Plan Date The planned date for the first execution of this test case.
Actual Date The actual date it was first run.
Comment Any comments related to the test case, required for those
test cases in a “Skip” or “Blocked” state.
Table 8: Test tracking
As the test organization runs each test, the state of each case will change from “In
Queue” to one of the other states noted in the table above. In the ideal situation, at the
System Test Phase Exit meeting, all the test cases will be in a “Pass” or “Closed” state on
the Test Case Summary worksheet for the final pass.
For each test that identifies a problem and enters a “Fail” or “Warn” state, the tester will
open a bug report in Tracker. For each defect, Tracker will store (at a minimum) the
information shown in the following table.
Field Meaning
Bug ID A unique identifier for each bug.
Title A one- or two-sentence summary of the failure observed.
Failure Description A text field, free-format, consisting of three sections:
Steps to Reproduce: A detailed, numbered process that
will recreate the bug.
Isolation: The steps performed to isolate the problem,
including verification on other platforms, bad-unit
checking, and other pertinent tests.
Regression: A short description of whether this failure is
a regression, and why or why not.
Severity The absolute severity of the failure mode in question,
without consideration of whether or not the failure mode
is likely in operation, on a scale of 1 (worse) to 5 (least
dangerous). The definitions are:

Field Meaning
1. Loss of data.
2. Loss of functionality.
3. Loss of functionality with a workaround.
4. Partial loss of functionality.
5. Cosmetic error.
Priority This is the priority, from a business perspective, of the
failure mode. In other words, assuming we have this
failure mode active or latent in our product on release,
how badly will this affect our ability to use this product
to assist “Some Client” in closing home equity loans and
lines of credit over the entirety of the expected product
life? Again, we use 1 (most dangerous) to 5 (least
dangerous). The definitions are:
1. Defect renders system useless; must fix for ship.
2. Defect will unacceptably impact revenue; must fix for
ship.
3. Fixing defect may not be as important as shipping
product on schedule.
4. Release date more important; fix in next release.
5. No financial impact expected; defer or fix at
convenience.
Resolution Notes Once the bug is inactive (see Activity field below,
“Closed”), this should include a description of the final
resolution.
Submitter The name of the tester or other engineer who identified
the problem, defaulting to the current user. For remotely
generated reports, this will specify either the contact
name or the outsource test organization itself.
Submit Date The date on which the bug report was opened.
Owner The person responsible for moving the bug report to its
next, and ultimately terminal, state.
Activity To capture whether any action is pending on a bug, this
field is either “Open” (further action is planned or needed
at this time) or “Closed” (no further action will occur for
Release 1.0). Open bug reports are in a “Review,”
“Reported,” “Assigned,” “Can’t Reproduce,” “Test,” or
“Not Resolved” state, while closed bug reports are in a
“Rejected,” “Defer” or “Resolved” state.
State The state of the issue, as follows:
Review: Awaiting a peer review by another tester.
Rejected: Review failed; behavior is not a bug.
Reported: The problem is deemed by the test engineer
fully characterized and isolated.
Assigned: The problem is accepted as fully characterized
and isolated by development, and an owner, responsible
for fixing the problem, is assigned.

Field Meaning
Test: Development have repaired the problem in some
level of hardware or software, and someone owns testing
the problem to evaluate the fix.
Not Resolved: The fix failed the retest.
Defer: Do not fix for this release.
Resolved: The fix passed the retest.
Symptom A classification of the symptom type as follows:
Functionality.
Performance.
Usability.
Reliability.
Operations.
Compliance.
Design.
Subsystem Affected A classification of the subsystem most impacted by the
bug as follows:
User Interface.
Bank Interface.
“Some Other Loan App”.
Scoring.
Other.
Release Status For old bug tracking database compatibility, from the
following list:
NDE, High.
NDE, Low.
IFP, High.
IFP, Low.
Deferred.
Must Fix For Phase A target phase for bug resolution from the following list:
Unit Test.
System Test.
UAT.
Deployment.
Beta.
Production.
Next Release.
N/A.
Test ID The test identifier (from the test-tracking spreadsheet
described above) corresponding to the test case the
engineer ran that uncovered the issue. Also allowed are
“Ad Hoc”, “Other” and “Unknown”.
Opened Count A numeric field incremented each time a bug is reopened,
to gauge the extent of regression.
Version Number The version of “Some Loan App” against which the bug

Field Meaning
was identified. (See “Release Management” section for a
discussion of version numbers.)
Closed Date The date on which the issue was confirmed fixed or put
on hold, which is used only when the issue is in a
“closed” or “deferred” state.
SIR ID Number If the bug was identified and reported using the old bug
tracking database, this is the old bug ID number for
cross-reference.
Table 9: Bug tracking
In order to ensure high-quality bug reports, the test team shall use the following process
for reporting bugs.
1. Structure. Testers shall use a deliberate, careful approach to testing, and take careful
notes. These notes form the basis of the failure description.
2. Reproduce. Testers shall check reproducibility of a failure before writing a bug
report. The steps to reproduce shall appear in the failure description as discussed
above. Sporadic failures shall be indicated by the keyword “intermittent” in the
summary and the failure description fields.
3. Isolate. Testers shall isolate defects by changing certain variables, including the CC
client tested, that may alter the symptom of the failure. The isolation attempts shall be
documented in the failure description fields
4. Generalize. After the tester has an isolated and reproducible case, he shall investigate
more general instances of the problem. Does the same failure occur in other modules
or locations? Can he find more severe occurrences of the same fault?
5. Compare. If the test case or test condition exposing the bug was run previously, the
tester shall indicate in the failure description field whether the bug is a regression.
Also, if the failure involves an invalid product offering, the test shall indicate in the
isolation section of the failure description field whether “Some Other Loan App”
exhibits the same bug.
6. Summarize. The tester shall write a one-line executive summary for each bug report,
in non-technical language, that explains how the failure observed will or might affect
the Call Center agent and/or Somebank’s business.
7. Condense. The tester shall eliminate extraneous steps or words.
8. Disambiguate. The tester shall ensure that readers will not misinterpret the report due
to words or phrases that are vague, misleading, or subjective.
9. Neutralize. The tester shall use only neutral, objective wording in their bug reports
and confine their statements to those of facts.
10. Review. The tester shall submit his or her bug report for peer review prior to
submitting them to the Development and Project Management teams.
The test team shall write an accurate, concise, thoroughly-edited, well-conceived, high-
quality bug reports. Comments, questions, or concerns about this process or the quality of
the bug reports coming from MTG during System Test should be directed to the Rex B.
To avoid test escapes, the testers shall adopt an active bias towards bug reporting. This
implies the following attitudes and behaviors:
• The Test Manager shall inform the team that its role is to find bugs, and shall
reinforce this message.

• If in doubt, testers are to assume the observed “Some Loan App” behavior is incorrect
until they satisfy themselves otherwise.
• If “Some Other Loan App” and the “Some Loan App” disagree about the correct
product offering, a bug exists and shall be reported.
• If the specifications, on-screen help, or any other official document and the “Some
Loan App” disagree about correct behavior, a bug exists and shall be reported.
• The tester shall report as a bug any event leading to loss of data or the crash, wedge,
hang, or other availability incident involving any server or network.
• If, in the professional opinion of the tester, the behavior of the “Some Loan App”
does not conform to reasonable expectations of Web and/or GUI program behavior
and/or quality, or is otherwise confusing, misleading, or ambiguous, the tester shall
report a bug.
• Questions about which component a bug resides in do not affect whether anomalous
behavior is incorrect. Testers shall report such bugs regardless of whether the
assignment of responsibility for repair is clear.
• Disagreements between MTG and any other group about whether a behavior is
incorrect shall be escalated to Magdy J. or Robert S. for resolution. They shall have
the sole authority to cancel or defer any bug report for whatever reason.
Release Management
During System Test, MTG will receive “Some Loan App” builds every Monday morning
from the Development teams. MTG does not anticipate any need for new “Some Other
Loan App” or Scoring builds into their respective Regression and QA regions, but, if
such occur, they will occur in sync with a “Some Loan App” build delivery. MTG will
run the entire set of tests against these builds. See Table 10 below for the build delivery
and test schedule.
Build Delivery Date Test Cycle/Test Pass Tests Run
4/23/98 Smoke All
4/27/98 1 All
5/4/98 2 All
5/11/98 3 (including agents) All (plus agent exploratory)
5/18/98 4 (including agents) All (plus agent exploratory)
5/25/98 5 (including. Agents) All (plus agent exploratory)
6/1/98 6/Validation All
Table 10: Release schedule and test cycles
The following five items are important for MTG in terms of release management.
1. Predictability and timing of releases. Releases that show up at unpredictable times,
too frequently, or too rarely can impede the forward progress of testing. Releases are
scheduled to move into the QA Region every Monday morning, with test launch at
10:00 AM. [Magdy J.: Can we convince SRM to support this schedule?] (When it is
necessary to accept an out-of-cycle release to keep testing moving, test will do so.)
Builds are moved into the Integration Region daily.

2. Update apply process. Ideally, the process of moving to a new release requires no
external support, is simple, and is executable (via automation) by the Development
teams in a matter of minutes. The specifics of this process for the “Some Loan App”
in the Integration and QA Regions are discussed below.
3. Update unapply process. Sometimes bug fixes create more problems than they
resolve, in which case a process for removing the update or recovering from the
change is needed. Like the update process, simplicity and automation are goals. The
specifics of this process for the “Some Loan App” in the Integration and QA Regions
are discussed below.
4. Naming. When the test team reports new bugs, they need a way of identifying the
offending releases. This requires a consistent naming convention for any subsystem
release to test. The naming convention need not be meaningful, but it should imply
sequence, as in A, B, C… For the “Some Loan App”, the releases imply both
sequence and date. They are of the form SLA.MM.DD.YY.rN, where MM.DD.YY is
the date of the build, and N is the build number for the day; e.g.,
SLA.03.07.00.r1.
5. Interrogation. Naming conventions do very little good unless a tester has a way of
determining the release names of all the subsystems by interrogating the system under
test. This process needs to be as simple as possible to enable quick reporting of bugs.
[Hanna R.: Please provide method.]
When the Development teams provide a build to the test organization, they will also
provide release notes. These release notes will, at a minimum, identify, by Bug ID, all
bugs fixed in that release. A bug is considered “fixed” from a development standpoint
when a developer has expended effort to resolve it and when he has run a unit test to
confirm that the problem is indeed resolved. The release notes shall be provided as a text
attachment to a release notification e-mail.
The release process to the Integration and QA Regions is as follows:
1. Webdotbank develops on Webdotbank-owned boxes.
2. Webdotbank performs pre-integration testing in Pre-Integration Region (hegdbqa2 -
224).
3. Some Loan App Development team develops on Developer’s workstations.
4. Some Loan App team performs Integration Testing in the Train Region.
5. Webdotbank (Max J./Jacques F.) shall perform a build each evening. (If there is no
build necessary for that day, an e-mail shall be sent to the parties list below informing
them of this.) Each build shall contain complete version information and file naming,
as well as a reference to the bug(s) fixed within that build and any associated DB
script changes or similar documentation. Each build shall also be supplied with
release notes as described above. Once the build is complete, Webdotbank shall
notify the following parties that a build has been completed and is ready for
promotion to Integration:
• Hanna R.
• Jenny V.
• Clarence T.
• Rex B.

• Jenna B.
• Emma G.
• Magdy J.
6. At 9:00 AM each weekday, the Build Manager (Clarence T.) shall promote the
nightly build to the Integration Region. If any Oracle database changes are necessary,
Clarence shall work with Minnie P. in SRM (our DBA) to make those changes.
7. The Development Team shall perform Smoke Testing (limited Regression Testing) in
the Integration Region.
8. Friday afternoon, Clarence T. shall inform the “Some Client” SRM contact, Petra V.,
of the revision of the “Some Loan App” to be promoted to the QA Region
(nclqa.fusa.com ) and Production Region (hegprod.bank.com ) on Sunday. On
Monday morning, Clarence T. shall confirm that the build was successfully promoted,
then shall notify the following people via e-mail:
• Magdy J.
• Rex B.
• Emma G.
• Harry W.
• Jenna B.
• Hanna R.
9. Should the quality of the new build fail to meet the continuation criteria described
above, the Test Manager shall halt testing and ask the “Some Loan App” Release
Manager to perform an “unapply” operation, reverting to the previous release. Testing
shall resume against that release, while waiting for an out-of-cycle build to repair the
blocking problems.
The Scoring QA Region and “Some Other Loan App” Regression Region shall not
change during “Some Loan App” System Test, except as the result of bugs found in
Scoring or “Some Other Loan App” during “Some Loan App” System Test. If bug fixes
or other code changes must occur in these regions to address such bugs, then the “Some
Other Loan App” Release Manager and Scoring Release Manager shall coordinate with
the “Some Loan App” Release Manager to ensure that all necessary code changes are
applied to all three regions (“Some Loan App” QA Region, Scoring QA Region, and
“Some Other Loan App” Regression Region) as part of the weekly build install.
Risks and Contingencies
The following table describes the key risks to success of this plan, and contingencies to
address them.
Risk Contingency
Insufficient test lab hardware. Consider a day and a night shift, weekend
hours, and so forth.
Tracker not available. Test team enters all bugs by hand. This is a
very inefficient and undesirable
approach; I estimate a 10 to 25 percent
hit.
Release management not well-defined,
resulting in a test cycle’s results being
invalidated.
Define a crisp release management process.
Weekly builds into the QA Region only
during System Test proper.

Risk Contingency
QA Region not ready for testing, leading to
results not necessarily representative of
the Production Region’s expected
behavior.
Start testing (functional only) in the
Integration Region until the QA Region
is ready.
Slip the System Test schedule day-for-day
until the QA Region is available.
Test lab setup delayed or incomplete. We can begin testing if the cloning is not
done, but, if we must function with
insufficient hardware (less than
described in this plan), we will suffer a
schedule impact.
Test environment system administration
support not available or proficient.
Identify system administration resources
with pager/cell 24x7availability and
appropriate Unix, NT, and network
skills.
Buggy deliverables impede testing
progress.
Complete unit testing by Webdotbank.
Adherence to test entry and exit criteria.
Test and/or product features, quality, or
schedule plans change, causing a
rethinking of the test project.
Change management or change control
board.
Expansions in test system scope may
necessitate “splitting” the test passes
over multiple cycles.
“Some Other Loan App” or Scoring change
without coordination with “Some Loan
App” releases.
The “Some Loan App” is closely tied to
these systems, so such changes reset the
current test cycle and require starting
over. A one-week slip in the System Test
Phase Exit date should be expected
should such an event occur.
Table 11: Risks and contingencies
Change History
The following table outlines the change history for this document.
Revision Released Description/Changes Author/Editor
0.1 3/26/98 First draft with questions and feedback
requests to “Some Client” team.
Rex B.
0.2 3/29/98 Second draft with responses to some of
the questions incorporated, sent to a
wider audience.
Rex B.
0.3 4/1/98 Minor changes based on feedback from
test team.
Rex B.
0.9 4/10/98 Incorporated comments from the plan
review. Remaining questions escalated
for resolution.
Rex B.
1.0 TBD Final Rex B.
Table 12: Change history

Referenced Documents
See the various documents in “SomeNetworkRepository” for more information on how
Release 1.0 of the “Some Loan App” is supposed to work.
See Julis R.’s performance document for more information on performance testing.
This test plan format complies with ANSI/IEEE Standard 829.

Frequently Asked Question
The following section addresses some frequently asked questions, and should be kept as a
reference for MTG team members and those who interface with them.
• What do I do if I get stuck and can’t run tests?
Good question. If you are stuck because you can’t figure out what the “Some Loan
App” should be doing, and you have checked “Some Other Loan App”’s behavior
already, contact Hanna R. and Jenna B. If you are stuck because the infrastructure is
misbehaving, contact Clarence T. or SRM as specified in the “Escalation Process”
section.
• How do I get in touch with people when I’m stuck?
The escalation and release processes described in this test plan are the normal
channels to be tried first. Phone numbers are included in the contact lists. Try normal
channels before escalating outside the process. If you can’t get the issue resolved
through normal channels for whatever reason, call Rex B. on his mobile phone at +1
(768) 555-1567. If he doesn’t answer, call him at his home office at +1 (176) 555-
1730. If you still can’t get Rex, call Magdy J. on his mobile phone at +1 (779) 555-
1540, or, if you can’t reach her there, use her pager, +1 (888) 555-1984. DO NOT
STAY STUCK. DO NOT GIVE UP. CONTINUE TO ESCALATE UNTIL
SOMEONE TAKES OWNERSHIP OF GETTING YOU UNSTUCK.
• Do I have to file a bug report every time I encounter a problem?
Yes, usually. If a bug report has already been written—by you or by someone else—
you should add a note. If the problem is an infrastructure problem—e.g., the QA
Region is very slow, a workstation is crashing, etc.—file a bug report and then
proceed as described in the “Escalation Process” section.
• Do I need to have all my bug reports reviewed?
Yes. All of them. Every single one. No bug report is to be formally submitted without
a review. If you can’t find anyone, call Rex B. on his cell phone and read the bug
report to him.
• What if I’m confused about how PVCS Tracker works?
Emma G. is the resident expert and the informal Tracker support resource for MTG.
Ask him.
• How do I know what version of the “Some Loan App” is running in which region?
Ideally, you could ask the “Some Loan App”, analogous to the “Help/About” options
in Windows applications. Right now, you can’t, so see the paper taped to the left wall
(the one with the Tracker workstations against it).
• When do new builds show up?
The QA Region is to receive a new build every Monday morning from SRM. This
process is still in flux, so Jenna B. will need to drive the process manually until it
begins to work regularly. The Integration Region is to receive a new build every
morning, given that Webdotbank has made a nightly build.
• How do I know whether my bugs are fixed in the new build?
The release notes for each nightly build specify the bugs fixed in that build. So, in the

QA Region, the complete set of release notes for all builds since the last build
installed let you know which bugs are fixed. As a practical matter, Jenna B. will print
a report every Monday for the bugs that are assigned to him and a “Test” state, ready
for confirmation testing in the QA Region. The appropriate MTG team member or
Developer will test each bug.
[Jenna B./Emma G., what do I need to add here?]

Case study-loan-processing-test-plan

More Related Content

What's hot (20)

Similar to Case study-loan-processing-test-plan (20)

Recently uploaded (20)

Case study-loan-processing-test-plan