Data Migration in Honduras

1
Data Migration in Honduras

• Challenges and Approach to Data Migration from M2
  to MIFOS

2
Overview

• The Adelante Foundation
• Adelantes Existing System
• Key factors in choosing Mifos
• Choosing a Migration Scenario
• Challenges in the Migration
• Attacking the Challenge
• Conclusions

3
Where is Honduras?
4
About Adelante Foundation

• Non-regulated private development organization
• Spanish-speaking clients and staff
• Founded in 1999 after Hurricane Mitch
• 5,000 current clients
• Strongly based on the Grameen model
• Growth limited by existing system

5
Adelantes Mission
Improve the standard of living among the
extremely poor women living in Honduras
Assembly of Adelante clients in Tocoa, Honduras
6
More Pictures
Signing contracts at an assembly
Many clients sign with an X
7
Adelantes Existing System

• M2 by Southern Horizon Software with heavy local
  customizations
• Microsoft Access database back-end
• Commercial, closed-source program
• No remote access capability
• Heavy reliance on paper forms
• Many manual calculations and repeated entry of
  data on different screens introduces errors

8
Key Factors in choosing Mifos

• Mifos meets clients needs better than any
  off-the-shelf system
• Mifos is developed by Grameen
• Mifos has integrated remote access data entry
  can occur closer to the data source
• Mifos is more flexible and extensible because
  its open source

9
Choosing a Migration Scheme

• Three options considered
• Client data and savings balances migrated only
  new loans entered into MIFOS
• All loan data from Jan 1, 2007 to present
• All loan data since M2 was started (2003)
• Options 2, 3 rejected because of additional time
  required and major differences in database
  schemas between M2 and MIFOS
• Option 1 maximizes likelihood of successful MIFOS
  implementation
• Caveat Option 1 requires legacy system for
  reports for about one extra year

10
Challenges in Migration

• Evaluate quality of source data via profiling
• Map data fields from M2 into MIFOS
• Perform and test automated migration
• Migrate in stages by branch office during a
  rolling go-live
• Ensure data entered into MIFOS has the same or
  better quality than data entered into legacy
  system

11
Sample of Data Mapping
12
Considerations in Mapping

• One-to-many mappings among tables
• Different values (e.g. dependents vs children)
• Different data types (e.g. integer vs char)
• Required fields in Mifos missing in source data
• Attention paid to constraints and primary key
  issues (e.g. Client_ID auto-incrementing in Mifos)

13
Attacking the Challenge

• Carefully map fields between M2 and MIFOS
• Note any calculations/transformations needed
• Data validation at every step
• Migrate M2 data (MS Access) to MySQL using
  Navicat
• Perl scripts to perform data mapping
• Read and write from MySQL directly
• Perl DBI interface to data
• Multiple iterations until no errors raised

14
Conclusions

• Like many MFIs, Adelantes existing system is
  inadequate and limits the organizations growth
• Mifos meets the clients needs and is considered
  risk-worthy because of the Grameen assocation
• Data migration is a critical piece to a
  successful implementation of Mifos at Adelante
• The flexibility of the open-source model provides
  a unique advantage over competing off-the-shelf
  commercial packages