Current Status

Introduction

Data Migration is typically about taking data from a source system and getting it into a destination system. This can be achieved manually, or through automation or some mix of both.

Data migration typically involves:

Migrating to Mifos

In relation to mifos, data migration is used to get a customer off their existing MIS approach and onto mifos so they may use the mifos application.

There is an existing deployment methodology which involves a data migration stage of which three possible approaches are outlined:

  1. No Data Migration
  2. Opening Balances Only
  3. All Historical Data

No Data Migration

Opening Balances Only

Employing this approaches means dropping all historical data and as a result the Datawarehouse and historical based reports are not useful for these accounts out of the box.

As mifos doesn't really support opening balance for loan/savings accounts, and there might be issues around 'adjustments of last transactions' as mifos at present can make assumptions about the first transaction being a disbursement (with respect to loans)

If opening balance approach is useful and used often, we should build this capability into mifos domain model and services and as such when migrating data into mifos, avoid possible problems with other behaviour in the system.

All Historical Data

As mifos (through an API or in model) doesn't really support the creation of loans, savings, client etc with historical data; Employing scripts that go straight to the database schema have being the only way of achieving something like this.

We should build this capability into mifos domain model and services and as such when migrating data into mifos, avoid possible problems with other behaviour in the system.

Analysis of approaches used in past

It would be useful to track what approaches have being used in the past and why and any problems encountered while analysing source data/application (see Existing Data Migration Tips/Problems for some examples of issues). Ryan Whitney gave a useful overview of the data migration process that was employed for a customer recently and what approach they went with and showed what they used to implement this also.

MFI

Size/Tier

Organisational Details

Approach Desired

Approache Used

Notes

Customer X

Tier 3

~100 office/branches

unknown

Opening Balances

see Ejemplo de proyecto de migración de datos

From keeping track of past and current migration details and we can better understand how to add value to mifos to make migrations a little easier and consistent hopefully.

Some Opinions

In Existing Data Migration Tips/Problems some issues were noted that caused problems. Some of the issues were around conceptual differences between source and destination systems and as such source may not have information for putting into destination and vice versa, mifos doesn't support a concept that is allowed in source system. If this concept was common enough in microfinance then would it be fair to say that mifos should try to model/support this concept?

Other Sources/links

Existing Data Migration Strategy

Existing Data Migration Tips/Problems

Tejus Data Migration Strategy

Data Migration Tools (Legacy I Think)