The Secret Life of AWS: The Time Machine (AWS Backup)

#aws#backup#devops#cloud

Code is replaceable. Data is not. How to protect your business with AWS Backup.

Part 28 of The Secret Life of AWS

Timothy was whistling as he typed. He felt invincible.

"I am just running a cleanup script," he told Margaret, who was walking by with her morning tea. "I need to delete the test orders from the Dev environment so I can run a fresh simulation."

He hit Enter.

DELETE /tables/Orders/items WHERE status = 'TEST'

His terminal spun for a moment, then returned a success message. Timothy smiled—until he looked at his other monitor.

The dashboard for his Production application suddenly flatlined. The "Total Orders" count dropped to zero.

Timothy stopped whistling. The blood drained from his face.

"Margaret," he whispered, his voice trembling. "I think... I think I just deleted the production database."

Margaret set her tea down and looked at the screen. She didn't panic. She didn't scold him. She just pulled up a chair.

"It happens to everyone, Timothy," she said softly. "You likely had your AWS CLI pointing to the wrong profile. Breathe."

"I can redeploy it!" Timothy stammered. "I have the SAM template from yesterday! I can run sam deploy and get the table back!"

"That will bring back the Table," Margaret corrected him gently. "But it will not bring back the Names. It will not bring back the Addresses."

"Your Blueprint (Episode 27) can replace your infrastructure," she explained. "But nothing in your code can replace your data. Once it is gone, it is gone."

"Unless," she added with a small smile, "you have a Time Machine."

The Backup Plan

Margaret navigated to the AWS Backup console.

"In the old days," she said, "we had to write scripts to back up every database individually. But AWS Backup is a centralized command center that manages protection for almost everything you build."

She clicked on Protected Resources. There, sitting in the list, was OrdersTable-Prod.

"You... you turned it on?" Timothy asked, hope returning to his eyes.

"I set up a Backup Plan last week," Margaret nodded. "Think of it as an insurance policy. I told AWS: 'Take a snapshot of this database every night at 3 AM, and keep it for 30 days'."

The Point in Time

"But the nightly backup was at 3 AM," Timothy said, checking his watch. "It's 10 AM now. If we use that snapshot, I've lost seven hours of orders. That's hundreds of customers."

"You are right," Margaret agreed. "If we relied only on the nightly snapshot, those seven hours would be gone forever."

"But for critical databases like DynamoDB," she continued, "we enabled something more powerful."

She pointed to a setting: Point-in-Time Recovery (PITR).

"This feature doesn't just take snapshots," she explained. "It watches every single change you make. Every write, every delete. It records the history of your data, second by second."

"It costs a little more than standard backups," she warned, "but when you need to save the business, it is worth every penny. It lets us go back..."

"To any second I want," Timothy realized.

The Restoration

Margaret clicked Restore.

The console asked for a time. Margaret looked at the logs. "You ran your script at 10:04 AM. So let's go back to 10:03 AM."

She entered the time and clicked Restore.

AWS began to work. It didn't touch the broken table. Instead, it created a brand new table called OrdersTable-Restored, populated with the data exactly as it existed one minute before Timothy's mistake.

Timothy watched the "Total Orders" count on the new table. It wasn't zero. It was 15,420.

"They are back," Timothy exhaled, slumping in his chair. "Every single one of them."

The Safety Net

"You have learned a scary lesson today," Margaret said, patting his shoulder. "Infrastructure as Code is powerful. It lets you rebuild your house in minutes."

"But data..." Timothy said, looking at the restored table. "Data is the life inside the house."

"Exactly," Margaret smiled. "You can 3D print the walls. But you cannot 3D print the memories."

Timothy pointed his application to the new table. The dashboard lit up green. The business was back online.

He opened his template.yaml file. He wasn't just going to rely on Margaret's console settings anymore. He was going to define his safety net in code: PointInTimeRecoveryEnabled: true.

He would never build anything without a Time Machine again.

---

Key Concepts

• AWS Backup: A centralized service to automate data protection across AWS services (DynamoDB, S3, RDS, EBS, etc.).
• Infrastructure vs. Data: The critical distinction that IaC tools (SAM) restore resources, but Backup tools restore content.
• RPO (Recovery Point Objective): The maximum amount of data (measured in time) you can afford to lose. (e.g., "I can lose 1 hour of data").
• Point-in-Time Recovery (PITR): A continuous backup feature that allows you to restore data to a specific second, minimizing data loss to almost zero.

---

Aaron Rose is a software engineer and technology writer at tech-reader.blog and the author of Think Like a Genius.