Building a Data Foundation Pillar #2: Build the Data Architecture

In our last article, we talked about defining your data strategy as a critical step for building a solid data foundation. We’ll now discuss the second, yet no less important, step: articulating the data architecture that will help make your data strategy a reality.

 
Data architecture is a high-level description of how information flows from place to place within your organization. It may include models, diagrams, policies, and schemas for collecting, storing, and analyzing your enterprise data.

In this light, it’s clear why defining your data architecture should only be done after defining your overall data strategy. For example, there are a variety of data repository options at your disposal—from enterprise-class data warehouses and smaller data marts, to operational data stores and data lakes. Each one of these options has a different function and use case.

Deciding how  to store your data, which is the central question of data architecture, can only be done once you know what data to store, why you’re storing it, and how you plan to use it within the organization—which are all foundational questions of data strategy.

Beyond these different storage options (and how information will flow in and out of them), there are other basic data architecture questions that you need to answer. For example, will your compute and storage resources be in the cloud, on-premises, or using a hybrid solution that gives you the best of both worlds?

Finally, data architecture also encompasses issues of data modeling and data taxonomy. Data modeling is the practice of creating conceptual models for how information moves throughout your organization. This includes:

• Diagrams that represent the relationship between different data objects.
• Database schemas that offer high-level blueprints of how information is stored within a structured database.
• Metadata objects and tables for storing contextual information about your enterprise data. For example: who created a given data record, when was it created, what is its privacy level, and how is it used throughout the organization?

 
Data taxonomy is the practice of classifying your enterprise data into relevant categories, providing a unified view and common terminologies across the organization. When knowledge workers spend an average of 15 to 35 percent of their time searching for information, better data taxonomy is a critical task. Your data can be structured in many different ways and sliced and diced across multiple axes, including:

• Geography
• Lines of business
• Marketing channel
• Time and date
• Location

 
With so many interlocking concerns, setting the data architecture is one of the most important steps when building a rock-solid data foundation. However, it’s far from the only one. Can’t wait to find out about all four steps? Download our white paper “Build a Data Foundation on Trusted Data.”