mdm and metadata management on a shoe string

16
MDM and Metadata Management on a Shoe String

Upload: others

Post on 18-Mar-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

MDM and Metadata Management on a Shoe

String

2

Colorado Housing & Finance Authority

Mission is to finance the places where people live and work in Colorado.Primary business is to make loans to low and moderate income homebuyers, affordable multifamily rental housing developers, and small and medium sized businesses. $3.4 Billion Portfolio / 30,000 loansProvides education and technical assistance about affordable housing and economic development.Does this through a network of private, nonprofit, and public partners such as banks, developers, and local governments.

3

Agenda

BI at CHFAArchitectureOrigin of MDMSolution RequirementsMetadata Data ModelETL ControllerCHFAPedia

4

BI at CHFA

Small set of frequently revised KPI’s drive businessBI to the rescue during market disruptionSingle Family acquisition and servicing

Current solutions are quick-win deliverables

Components of enterprise BI solution are in place

CHFA seeing value in common business terms and definitions

BI Target Architecture

5

Transactional Systems

JDE

NLSOrbit Tax Credit HDSFidelity

ADP

ISC

Insight REMS

Stars

External Source of Funds

Comet

Integration LayerBy subject area• Corporate definitions• Fields• KPI’s• Dimension conformity• Integrated base data• Data quality• History repository

Data Warehouse

CHFAPedia

Enterprise

Future BI Environment

Presentation Layer

ExcelDashboards

Data Delivery• Dashboards• Excel• Reports Reports

Subject LayerStar Schemas by subject area or business need• Easy to query• Providing history• Validated• Referenced metrics• Referenced analytics

Staging Layer

• Consolidated data• Exact replica

Staging Database

By application

6

Origins – Servicing ProjectActive project – marry loan origination and servicing data for life of loan analysis

Servicing system is outsourced

Business perspective

13 different definitions of “delinquency” =

Unhappy and confused executives

7

Origins – Servicing Project350 tables with 7500 fields feeding 400 reports and ad hoc queries =

1 Sad Business Analyst

8

Origins – Servicing ProjectArchitect Perspective

Moving from data mart to warehouse

Building standardized layers / stages to move from source systems to presentation layer

9

Solution RequirementsAs an executive, I want to get the same answer when I ask about our delinquency rate

As a business analyst, I want to keep track of reports, fields and definitions

As a general CHFA user, I want to go to one place to find business terms, reports, dashboards

As the BI architect, I want a uniform, scalable means for capturing, cleaning, transforming and presenting data

10

Metadata Project PhasesSimple MSAccess database as prototype

SQL Server, table-driven ETL with web-based control for data acquisition and transformation

Web-based application for business terms

Consolidated portal for reports and dashboards

Formal kick-off of data governance

Metadata Database Model

11

Division

Field

Business Term

Report

Database Table Value

Steward

ETL Steps

1212

Source Copy

Change Capture Clean Combine Data

Warehouse

CHFAPedia

Enterprise

Cube

T-SQL, Table Driven SSIS, Package Driven

Table driven, T-SQL optimal for easy transformations. It provides rich interaction and logging with MetaDataSSIS optimal for complicated transformations. Built in components provide free benefits (multi-threading, fuzzy lookup, automated notifications, etc.)Master SSIS package controls whole ETL process

Master

Metadata Dimensions Facts Cubes Report Warm Up

13

ETL Controller

14

15

16

Questions?

William [email protected]