SaaS Backup Procedures
This document explains the backup procedures followed by Research Space for customers of RSpace SaaS (software as a service), where provision of System Administration Services is included in their purchase.
On-Premise vs. SaaS
For on-premise deployments of RSpace, backup is solely the customer's responsibility. You can find some general advice about on-prem backups here. We will consult with your IT personnel at the time of deployment. For backing up AWS-based RSpace instances, Research Space uses scripts to automate the backup process that we are happy to share with customers on request. When deployed as SaaS, Research Space take full responsibility for the durability and accuracy of stored data.
Data Storage and Backup Frequency
Data is stored in a MySQL 5.7 or MariaDB10.3 database; files are stored unmodified on EBS volumes in a directory structure. The sorts of backups made are:
- We make hourly file syncs to S3 using AWS CLI tool
- Nightly and weekly snapshots of instances and data volumes are stored as machine images (AMIs). These are fast to make, and support RTOs in the order of minutes
- Logical database backups are made nightly, and stored on S3. Data Files, logs, configuration files and search indices are additionally synced to S3 hourly. This process ensures a completely separate, independent storage for research data, at the cost of longer recovery times. AWS S3 has extremely high durability (eleven 9s—see https://aws.amazon.com/s3/faqs/) and serves as a backup of last resort
Data backups are at the location of the AWS region selected by the customer at time of installation, or geographically closest to the customer, if no specific region has been designated.
What is included in the backup
All of these are backed up regularly (at least nightly):
- The MySQL/MariaDB RSpace database
- The RSpace filestore
- Log files
- RSpace configuration
Recovery of AMI snapshots is tested weekly using an automated procedure to ensure all data is recovered and database integrity is maintained, and that RSpace launches successfully.
The Disaster Recovery procedure from S3 backups is tested twice-yearly.
Retention & Deletion of Data
Data would remain in S3 backups and EBS snapshots until the backup expires (typically 3 weeks).
For all deployment options, specific requests to permanently delete individual items of data can be made on a case-by-case basis.
Customer Copy & API Options
If the customer has an AWS account, database and filestore backup AMIs and/or S3 buckets can be copied to your AWS account.
Additionally, we offer optional, redundant, automated / scheduled total data export options via our modern API to any repository or other destination of your choice.