When Moving Services to the Cloud, Think Data First

It’s true that agencies can use data to better support quick decision-making and to better serve their constituents. And it’s true that moving data to the cloud can improve their ability to use data. But there’s a catch: If your data’s a mess when you move it to the cloud, all you get is a cloud-based mess.

That’s the idea behind a data-led cloud migration. In moving applications and services to the cloud, you don’t want to move your existing data silos to the cloud. Instead, you want to centralize that data, making it easier to manage, secure and enrich.

“Just getting it to the cloud is … a good step, because you will reduce some of the costs that are associated with just ensuring the health of blinking green lights,” said Brian Schoepfle, a solutions architect at Amazon Web Services (AWS). “But you need to start leveraging the new capabilities that you will have by virtue of being in the AWS Cloud to really extract the kind of information that you want from the data that you have.”

Schoepfle highlighted the key steps in a data-led migration.

Focus on data quality, data quantity and data sources. Data can be extremely valuable, but only if it is free of errors and structured so users can interact with it efficiently. It’s equally important to have enough data to perform required tasks. In general, more data leads to more reliable models and better results.

Clean, enrich and transform data. Clean data is free of errors, with a high degree of integrity. After being cleansed, the next step is enriching the data by adding context from additional relevant sources. Finally, the data must be transformed, based on its intended use and the tools that will be used to interact with it.

Move data into cloud-based data lakes. Essentially, a data lake is a large pool of data. The goal is to centralize as much data as possible into data lakes, and to have as few data lakes or lakehouses (a combination of data lake and data warehouse) as possible.

Apply analytics, AI and machine learning. Whether it’s predicting outbreaks, uncovering fraud or improving public safety, good data is at the root. By taking the previous steps, analysts and models will have access to clean, relevant data to power insights.

This article is an excerpt from GovLoop’s guide “How to Provide People-Oriented Services: A Guide for State & Local Public Servants.”