In the era of digitalization, enterprise applications have been known to struggle to serve content from traditional file-based storage backends that run on local disk filesystems. With the number of files increasing exponentially, it can become cumbersome and difficult to manage the sheer volume of content on a local disk system within a server. 

Let's come back to our insurance company example. Now, before they can create a data lake of the historical contracts and risks involved, they need to capture that data somehow. As they want to capture as much historical data as possible, the company decides to scan and run optical character recognition (OCR) on all of their paper archives and unify that data into a new application. But the company's almost 100 years of records are going to consume massive amounts of data, as the images need to be scanned in at a high quality for the OCR process, and for the later archiving and replacement of the paper archive with a digital one.

The company already has an enterprise application that they have been using for a few years, where all contracts are now digital, indexed, and searchable. Now, they wish to add the OCR data and corresponding images to the new enterprise application. Adding the OCR data to their application proves to be simple, while adding hundreds of millions of files to the existing filesystem is determined to be too much for the on-premises application backend.

The images can be instead copied into S3 and the application can be extended by providing a link to the S3 image of the file when this file is required for review. This offloads the on-premises application from having to serve static content from an expensive enterprise-grade backend that was designed to run dynamic workflows such as the database that stores the index and content of the contracts.