MongoDB Backup and Restore on AWS: Automated Daily Backups to S3

Most people find out how good their MongoDB backups are on the worst day of the year. This guide covers the setup that avoids that on AWS: automated daily backups to S3, a restore you've actually tried, and an honest look at what daily backups don't cover. Still deciding whether to self-host at all? We priced MongoDB Atlas against self-hosting on AWS tier by tier.

Why MongoDB backups on AWS can't be optional

The failure that hurts isn't usually the server dying. AWS replaces hardware quietly. It's a bad migration script, a deleteMany with the wrong filter, or a compromised password. Replication copies those mistakes to every node within seconds, which is why a replica set is not a backup. Only a copy kept somewhere else, from before the mistake, gets the data back.

Regulation points the same way. GDPR and most security frameworks expect you to be able to restore personal data after an incident, and to prove you've tested it.

How the Meetrix MongoDB AMI backs up to S3

• Daily, automatic. The server writes a backup to S3 every day without anyone remembering to.
• Your bucket. You give an existing S3 bucket name when you launch the stack, so backups stay in your account, in a region you choose.
• Off the instance. If the EC2 instance or its volume is lost, the backups aren't.
• Standard tools. Nothing proprietary to restore from. The MongoDB Database Tools (mongodump, mongorestore) work against it.

Setting up MongoDB with automated S3 backup

You'll need an AWS account, basic familiarity with EC2 and CloudFormation, and an S3 bucket created in advance.

1. Open the MongoDB AMI on AWS Marketplace, click "Continue to Subscribe" and accept the terms.
2. Click "Continue to Configuration", pick your region, then "Continue to Launch" and choose "Launch CloudFormation".
3. Fill in the stack: a stack name, DeploymentName, InstanceType (t3.small is the recommended start), your key pair, MongodbDatabase, MongodbUser, a strong MongodbPassword, and the S3Bucket name for backups.
4. For SSHLocation and DatabaseSSHLocation, enter your own IP or your VPC range (for example 203.0.113.10/32), not 0.0.0.0/0. Leaving port 27017 open to the internet is how MongoDB databases end up ransomed.
5. Acknowledge that CloudFormation may create IAM resources, submit, and wait 5 to 10 minutes.
6. Take the PublicIp from the Outputs tab and connect on port 27017 with MongoDB Compass or mongosh, using the user and password you set.

MongoDB - Developer Guide

A comprehensive step-by-step guide to integrating MongoDB into your AWS environment using an AMI and CloudFormation, including setup, connection, upgrades, and troubleshooting.

By Dinesh Chathuranga • Meetrix.io

How to restore a MongoDB backup from S3

Practise this before you need it. The steps are short:

1. List the backups: aws s3 ls s3://your-bucket/ --recursive and pick the one from before the incident.
2. Copy it to the server: aws s3 cp s3://your-bucket/path/to/backup ./
3. Restore into a scratch database first and check the data with a few queries. mongorestore can rename namespaces on the way in with --nsFrom and --nsTo.
4. When you're sure, restore over production. Add --drop to replace collections instead of merging into them.

The flags depend on how the backup was written (a dump directory, or a single --archive file, often gzipped with --gzip). The mongorestore documentation covers each case. For more detail on the server side, see the MongoDB developer guide, or email Meetrix Support at support@meetrix.io.

What daily backups don't cover: point-in-time recovery

A daily backup taken at 02:00 can't bring back writes from 14:00. If losing up to a day of data is unacceptable, you need point-in-time recovery (PITR), and that means capturing the oplog continuously, not just taking dumps.

• Run MongoDB as a replica set, even a single-member one, so it has an oplog.
• Use a tool that streams the oplog to storage continuously, such as MongoDB Ops Manager or Atlas continuous cloud backup.
• Pair it with EBS snapshots for fast whole-volume recovery.

For most small and mid-sized apps, daily dumps to S3 plus nightly EBS snapshots are a reasonable trade. Be explicit about it, though: write down how much data you could lose, and make sure the business agrees.

MongoDB backup and restore best practices

MongoDB backup to S3: what it costs

Backups are cheap next to the database itself. S3 Standard is about $0.023 per GB per month in us-east-1, uploads from EC2 to S3 in the same region don't carry a transfer charge, and requests cost fractions of a cent. A lifecycle rule that moves backups older than 30 days to a Glacier storage class cuts that further. See S3 pricing for your region. The bigger line is the server: the Marketplace listing shows the software fee per instance type, on top of EC2.