 TestSentinel |
 |
|
Testing in
Migration Projects
Migration projects are frequently subject to significant cost
and delivery time overruns, and large projects are at significant risk of outright project
failure.
Why should this be so? The primary reason is that most projects from other vendors focus
on the changes required, and not on the testing required to assure that the work was done
accurately. Unfortunately, testing is not a well-implemented discipline. This is
particularly problematic because the cost of testing is roughly double the cost of
converting the programs: |
 |
The exact relationship
varies depending on technical and business factors, but the typical range for the cost of
testing is from 50% to 80% of the total cost of the project, or from a ratio of 1:1 to a
ratio of 4:1, with an average of roughly 2:1. If adequate test time and resources are not
planned into the project, it is at risk before it starts.
|
 |
The Forecross
methodology and supporting technology is designed to absolutely minimize these risks,
while increasing accuracy in the result and minimizing costs. Unique in the industry, our
methodology mitigates risk by automation of both the conversion and the highest cost
phases of testing. Automation introduces consistency, reduces costs, and improves
predictability and accuracy. Manual efforts produce the opposite. These effects become
more pronounced as the size of the project increases. |
The TestSentinel Solution |
TestSentinel is a capture/replay system that highly automates
equivalence testing for both batch and on-line programs. The basic methodology is to
capture all of the data input to and output from a program during execution prior to
migration, including the logic path taken through the program. The captured input data are
then replayed into the migrated program and the output data from the replay execution
automatically compared to the output data from the pre-migrated program. If there are
differences, a report is produced to assist a programmer in quickly identifying whether
the discrepancy represents a problem.
|
Instrumentation |
TestSentinel is based
on a source code instrumentation engine that inserts logically neutral code into programs
to automatically perform the testing functions without any requirement for human
intervention. This dramatically reduces the load on personnel, system and storage devices.
The source code delivered to Forecross in the inventory is downloaded to a PC platform
where the instrumentation process takes place.
|
Creation
of Baseline Capture File
After the original source has been instrumented and compiled,
the program is executed using either test or production data. All data input into the
program and output from the program are captured in a flat baseline capture file that is
used during the remainder of the testing process. This includes data from and to files,
databases, screens, printers, calls to the operating system and other programs, and any
other data entering or exiting the program.
|
Testing and Verification
After the program has been migrated, it is instrumented on a PC
for replay and code coverage analysis. During this instrumentation, code is inserted that:
1. Suppresses normal input operations;
2. Suppresses normal output operations;
3. Causes the program to obtain its input data from the baseline capture file;
4. Dynamically compares the program's output data against the output data in the baseline
capture file;
5. Reports comparison errors;
6. Provides coverage analysis.
The program that has been instrumented for replay is compiled and executed in replay mode.
During replay, the program obtains its input data from the baseline capture file instead
of normal sources and normal output operations are suppressed. The output data buffers are
automatically compared to the output data captured in the baseline capture file. If there
are differences, an exception report is produced. Once errors are corrected, the replay
test can be re-executed as often as desired without the need for a normal test environment
(as is required for conventional testing). |
| Testing
Mainframe Programs on a PC |
During replay, the
only requirements for the equivalence test are the ability to execute the program code and
the baseline capture file. There is no need to replicate test or production databases or
the normal operating environment for the program. In many cases such as IBM COBOL, there
is a PC version of the compiler, so replay can be performed on a PC at the Forecross
factory simply by downloading the source code and the baseline capture file.
|
| Code
Coverage Analysis |
During replay,
TestSentinel will compile a branch and path analysis coverage report that will show what
sections of the program logic have and have not been executed. If it is then determined
that the code coverage was insufficient, additional test data can then be prepared and
another capture and replay executed to ensure sufficient code coverage.
|
| Benefits: |
The capture/replay
process:
1. Provides an automated way to capture online and batch data during program execution
that is 100% accurate;
2. Can be executed using production or test data;
3. Provides an automated way to replay the captured data;
4. Automatically compares 100% of the data output from the program against the baseline;
5. Provides exception reports that quickly direct a programmer to the source of the error;
6. Provides a way to test a program without the need of its normal operating environment;
7. Provides the ability to repeat tests with no setup;
8. Provides the ability to unit test migrated programs at the Forecross factory.
The capture/replay process eliminates:
1. Manual identification of what files and databases a program uses;
2. Manual identification of the order the files are used;
3. Manual creation of test packs;
4. Errors associated with conventional gathering of baseline data;
5. The requirement for operators to precisely re-enter screen data;
6. The need for a file comparison utility;
7. The need for a screen capture/replay utility;
8. Human comparison of reports and screens.
|
Forecross is a registered trademark of Forecross Corporation.
Copyright © 1996-2008 Forecross
Corporation
All Rights Reserved. |
|
