IBM Cloud Transformation Advisor
Learnings from analyzing a legacy application
By Divakar R Mysore and Jagdish krishnan
Overview:
Migration and Modernization journeys to the cloud for a monolithic legacy application can be very complex considering the facts of variability like application design, implementation & deployment approach etc. Transforming or migrating legacy application to modern may require significant amount of time to analyze and lengthy assessment cycles required if a right Asset or Accelerator not considered.
In this article, we share our experience using IBM Cloud Transformation Advisor (TA) to analyze the application.
Application team objectives:
The monolithic legacy application considered for this article is a customer facing CRM application which was custom built using J2EE technologies along with enterprise frameworks like Struts, hibernate etc. deployed on IBM Application Server 8.1 having Linux as OS. This application is a home grown application that was built around 15 years ago and over time to time new features and modules has been added implemented using additional frameworks.
As part of the modernization journey, client had envisioned to move their current java workloads into Liberty on OpenShift container platform. We used TA to accelerate the application assessment and recommendations for moving the legacy application to OpenShift container platform.
Application current state:
•Application architecture and design documentation are very lean
• There were no enterprise standards that were consistently followed in the code
• There were more than 200 client or dependent libraries present in the class path of the application
• Apart from the shared library location the dependent libraries were scattered across in different locations like manifest files, server lib, JVM lib location
• There is no standard and modern DevSecOps process
List of technologies / frameworks used:
The selected application uses multiple J2EE technologies and frameworks in different modules as given below
• JSP & Servlets
• EJB 1.1
• Struts 2.0
• Hibernate 4.0
• JMS
• Other custom frameworks
Challenges with assessment
Complexity and size of application makes the assessment time consuming — Large number of dependent libraries present in the application class path that requires during application assessment for containerization
Dependent libraries that are placed outside the server location are referred by the application making dependency analysis complex
Transformation Advisor (TA):
IBM Cloud Transformation Advisor is one of the tool as part of IBM Cloud Pak for Application. TA provides the capability of assessment, strategic planning and execution during the modernization journey.
Please refer the below links for installation procedures and other details.
TA Installation using local setup
TA Installation on OpenShift
TA Overview
TA overview video link
TA Tour video link
TA application assessment for modernization video link
Our Approach:
We took a two-step approach to assess the applications. In the first step, we analyzed the application with the shared libraries in standard location. In the second step the results were augmented with the libraries in non-standard location.
First Step:
As a first step of our approach, based on the current client environment we have used Linux as source operating system, WebSphere application server as the domain and used “Application & configuration” for the analysis to retrieve the right data collector for the choices made.
To begin with, around 200 shared library dependent jars are assessed along with the main application ear using the below TA Data Collector execution script.
“./transformationadvisor-2.4.0/bin/transformationadvisor -w /opt/IBM/WebSphere/AppServer — scan-node — ignore-missing-shared-library — ignore-missing-binary -p AppSrv01 — applications Application1EAR”
For OpenShift Liberty container platform specific assessments and reports, Liberty runtime was selected as a “Migration target on cloud Pak for app”.
Results:
The assessment results for the application ear and its dependent shared libraries that were configured and mapped to the application successfully ran. The results were automatically uploaded to TA backend and all the three report namely Technology report, Inventory report and Analysis report was generated successfully.
Second Step:
Our second step in our approach is to get the TA assessments for the client dependent libraries or jars that are referenced in other locations like in the application manifest files, JVM libraries etc. To address this, we have used TA feature of using an argument that can scan the libraries outside of the shared library location.
With this approach TA Data Collector does not upload the assessment results zip of the libraries to TA back-end automatically as in the first step. So we uploaded the data using the upload option in TA.
Note: It should be noted that the same collection that was used to download the data collector should be selected to upload Assessment results zip. Else we may not have a holistic view in all the three assessment reports (Technology, Inventory & Analysis report)
Below is the Data collector script used to scan the client libraries outside the shared library location
“./transformationadvisor-2.4.0/bin/transformationadvisor -o /opt/CRM/Application1/lib”
Where the path “/opt/CRM/Application1/lib” is path of the client libraries outside the shared library location
Results:
Using step 2, we have successfully generated TA reports and loaded into TA using the upload data option. This gave a holistic view and the assessment reports considering the dependent libraries and modules referenced outside the shared library location.
Conclusion:
Using the options provided by the TA Data Collector as detailed in this article, our team has successfully assessed the monolithic legacy application by using the TA generated assessment reports for technology, inventory and analytics. Using these reports, we have identified the critical migration issues holistically for the applications and its dependent modules, complexity, refactoring efforts along with recommendations provided. Without using TA, It would have been very time consuming and error prone to analyze such a large monolithic legacy application.
