pilot verification and future road mapping for a user centric data warehouse application sundaresa...
TRANSCRIPT
Pilot Verification and future road mapping for a user centric Data
warehouse application
Sundaresa SubramanianHardik Patel
Nikunj Damani
Infosys Limited (NASDAQ: INFY)
Abstract
Financial institutions rely on reporting as a key indicator – both for operational efficiency and to monitor their health status. BFSI in the near past was talking about ‘just in time’ reporting – reports that could be generated quickly based on the needs and produced. BFSI in the current mode is all about ‘slice and dice’ reporting – intelligent reporting mechanism that not only provides you information but gives various perspectives to it.
2
Reports that can be sliced and diced. Merely being ‘on time’ is not enough in reporting today
Need for conference room piloting strategy?
Banks have to redesign or transform their existing data warehouse system to something more ‘efficient’. By ‘efficient’ we mean to say that the new solution should take ‘optimal time’ to perform a business function, consume ‘optimal cost’ and realize ‘optimal business benefits’.
Time, Cost and Benefit are always in a Mexican-standoff with one compromising the other if proper planning is not done for the entire program.
The best way to approach these type of futuristic programs is to adopt a “piloting” strategy where a set of critical business functions are implemented on a pilot basis for the transformation project.
This type of Conference room piloting requires agility, future vision, overall domain knowledge and testing acumen from the QA team since they act indirectly as architects for the holistic solution.
Through this paper we intend to discuss best practices from QA effort for a pilot data warehouse reporting and transformation program for regulatory compliance domain of a Bank.
3
Background
The primary aim of data warehousing is to make business-worthy data available as quickly and as accurately as possible. Let us consider the example of a complex data warehouse system for risk and compliance LOB of a financial institution. The typical reporting can be viewed in 4 different types as shown in Figure 2: Management reporting for strategic decision makingBusiness user specific reportsRegulatory compliance – External and internal auditsCapital adequacy specific reports – such as Basel II & III
4
Challenges faced in transformation projects There are multiple challenges involved in a typical transformation project of such magnitude where a legacy data warehouse application is transformed to a modern technology/architecture. Let us analyze those challenges/problems from various points of view:
6
View point Challenges while implementing a new solution (replacing legacy)Project Sponsor view Risk of failure
Budget constraints
Realization of strategic goalsTechnology Manager view
Architectural implications because of new technology
Known application Vs unknown technology promising greater thingsApplication Manager View
Seamless integration of systems with one another in new technology
Training and skill-set constraints
Increased upstream dependency (to provide source files for existing application as well as new development)
Environment management team
Resource constraints (System resources)
Performance constraintsTest Manager view End to End Test strategy for new technology
Realize ROI, improve productivity while testing new technology (there will be pressure on test team to utilize their legacy system knowledge and show productivity)
Business user view Skeptical about the new technology and its versatility in fulfilling the business needs
Why this Paper?
Conference room piloting helps eradicate the ‘fear of unknown’ while transforming to a new technology. It allows the project team to sample (or) pilot the new technology on a specific module or business flow first. The output is analyzed and lessons learned out of this conference room piloting exercise are explained to the business user.
In CRP, the business user plays a quick and early role in determining the success of the new technology by providing feedback based on the pilot.
7
Traditional data warehouse transformation projects – Big Bang
8
Figure 3: Traditional data warehouse transformation projects
Demo of new technology
High level transformation
roadmap
Architecture Level changes
Design ChangesCoding & BuildSystem Testing
User Acceptance testing
Error identification points
Conference Room piloting – Sampling the solution first…
9
Figure 4: Conference Room Piloting
Demo of new technology
Sample business flow or critical scenario identified for
transformation
Development & mini-build
Flash testingUser review of the
pilot transformationSignoff
Transformation roadmap based on lessons learned &
best practice from pilot
Error identification pointsTraditional development cycle
Conference Room piloting with early QA involvement
Though conference room piloting would provide everyone a sneak peak of new technology,
QA was not involved from very early stages in the lifecycle. If the QA team has good
knowledge on the existing application then their domain and technical acumen could be used
as valuable inputs during conference room piloting of a module. This gives rise to the concept
of early QA in conference room piloting, where the QA team would be involved in every stage
of the conference room pilot activity adding value to the entire process.
10
Application of early QA in conference room piloting
Existing data warehouse gets its data from 80 different source systems – about 40% of the systems are sourced directly and another 50% of the source systems pass through intermediate stages and roughly 10% of source systems have a combination – few data elements are sourced directly whereas few other data elements are passed through intermediate stages. Flow is illustrated in Figure 4. Various lines of businesses such as Mortgage, Reverse Mortgage, Auto Loans, Personal Loans, Aircraft/Boat loans and student loans are served by these source systems. The data warehouse is revamped (new database, new reporting routine). There is a need to identify one or two source systems for Conference room piloting purpose and the systems have to be selected from a group of 80 systems.
Objective selection criterion1
11
Early QA involvement - Objective selection criterion
Data warehouse
Data warehouseData warehouse
Direct sourcing Indirect sourcing Direct + Indirect
Intermediate DB
Source systems
Source systems
Figure 5: Various ways of sourcing data to data warehouse
Criticality of the source systems – based on defectsCoverage of all business flows – example Mortgage and LoansPrioritization based on type of sourcing - both direct and indirect sourcing
12
Application of early QA in conference room piloting
For an existing data warehouse program, a request has to be placed to the upstream to send test file for every iteration of development, system testing or UAT. Upstream followed its own release cycle and calendar and so test data requirements had to be communicated well in advance. The problem faced due to this scenario was that any adhoc test request was not honored by the upstream. Moreover, the upstream data sometimes did not cover all the test scenarios as defined by the QA team. When the data warehouse program was transformed and redesigned to a new technology, one of the critical requirements during conference room piloting was to decide a test data strategy that minimizes upstream dependency.
Minimize upstream dependency2
13
Early QA Involvement - Minimize upstream dependency
Existing data warehouse’s production data is analyzed
QA team considers this production data as a ‘baseline’ for
implementation of this new transformation program.
QA team prepares a test strategy to extract this production
data to a temporary database, process it and use it as “Golden
copy reference’ for any new test data requirements..
Since upstream dependency for test data is minimized because
of this strategy, the project team readily approves this suggestion
and starts implementing it.
14
Advantages of early QA in conference room piloting
Early involvement of QA team during CRP helped in their better understanding of the target architecture and hence led to effective test approach, planning and execution.
By following CRP approach, fewer defects were identified during the actual QA verification as the critical defects were identified and fixed during the POC.
There were no design gaps identified as each module POC was certified by QA team during CRP.
A significant cost saving was achieved in the project by minimizing upstream dependency.
Business stakeholders were very confident on the quality of the application due to exhaustive testing performed by QA in CRP.
15
Conclusion – Revisiting the objectives
View point Challenges How Challenges are mitigated in CRP + Early QA
Project Sponsor view
Risk of failure
Budget constraints
Realization of strategic goals
Cost of poor quality decreases considerably due to early detection of defects.
Increases predictability of the application
Technology Manager view
Architectural implications because of new technology
Known application Vs unknown technology promising greater things
Defects found after conference room piloting phase expose the architectural limitations of new technology (if any), enabling tech team to take informed decision on future roadmap.
Application Manager View
Seamless integration of systems with one another in new technology
Training and skill-set constraints
Increased upstream dependency (to provide source files for existing application as well as new development)
End to End QA vision and roadmap developed during CRP to uncover early integration errors.
Test data governance (by using existing production data) decreases upstream dependency
16
Revisiting the objectives…continued
View point Challenges How Challenges are mitigated in CRP + Early QA
Environment management team
Resource constraints (System resources)
Performance constraints
Early performance profiling during pilot or CRP phase resulting in bottleneck identification & resolution
Test Manager view
Realize ROI, improve productivity while testing new technology (there will be pressure on test team to utilize their legacy system knowledge and show productivity)
Suggest an automation roadmap and develop reuse during CRP – by leveraging existing domain and application knowledge of QA team.
Business user view
Skeptical about the new technology and its versatility in fulfilling business needs
QA team provides unbiased end user view of new technology.
17
References
Infosys project experience Infosys resources (www.infosys.com) www.wikipedia.org
19