Security policy · BCP-12
Backup, Business Continuity and Disaster Recovery Policy
1. Purpose and scope
This policy sets how Enigmatech Mexico SAPI de CV ("Enigmatech") backs up the EnigmaSuite Platform and restores it after a failure, a security incident or a disaster, so customers can keep operating and their data, including data received from partner platforms such as Amazon, remains available and protected. It applies to the production database, file storage, application servers, configuration and secrets. It supports the Information Security Policy.
2. What is backed up
| Component | Backup | Frequency | Retention |
|---|---|---|---|
| Production database (PostgreSQL, managed provider) | Provider-managed backups, encrypted at rest with AES-256 | Daily, with continuous write-ahead-log archiving where enabled | Rotated within 35 days |
| Geographically separate copy of the database | Encrypted copy stored in a cloud region different from the primary region | Daily | Rotated within 35 days |
| File storage | Provider redundancy across separate facilities | Continuous | Follows the record it belongs to |
| Application code and infrastructure configuration | Version-controlled repositories, with the full history | Every change | Permanent |
| Application releases | The previous release is kept on the server before each deployment | Every deployment | Until the next deployment |
| Secrets | Managed secrets vault and the providers' own redundancy | Every change | Until rotated |
All backups are encrypted, are reachable only by the named engineers who run production, and are never copied to endpoints or non-production environments.
3. Recovery objectives
| Scenario | Recovery time objective (RTO) | Recovery point objective (RPO) |
|---|---|---|
| Failed deployment | 15 minutes (automatic rollback) | 0 |
| Loss of the application server | 4 hours | 0 (no customer data lives only on the application server) |
| Corruption or loss of database data | 8 hours | 24 hours, or minutes where continuous archiving is enabled |
| Loss of the primary cloud region | 24 hours, restoring from the geographically separate copy | 24 hours |
Access to buyer personal information needed to ship pending orders is restored first.
4. Disaster recovery procedure
- The Incident Commander declares the disaster under the Incident Response Policy and names the recovery team.
- The team decides the recovery path: roll back, rebuild the application server from code and configuration, restore the database to a point in time, or restore in another region from the geographically separate copy.
- The restore is made to a clean environment, its integrity is checked against record counts and checksums, and deletions made after the backup (for example, buyer personal information past its retention period) are applied again before the system returns to service.
- Secrets are rotated if the cause could have exposed them.
- Customers are kept informed through the Platform and by email, and platform partners are notified as their terms require.
- A post-incident review records the timeline, the objectives met and the improvements needed.
5. Testing
- A restore of the production database to an isolated environment is tested at least every quarter, including at least once a year from the geographically separate copy. The test measures the time taken and the data recovered against the objectives above, and the results are recorded.
- Rollback of a deployment is exercised automatically by every production deployment.
- The disaster recovery procedure is walked through at least once a year as part of the incident response tabletop exercise.
6. Business continuity
- Critical services (database, authentication, file storage, edge network) run on providers with redundant facilities and independent certifications.
- Runbooks for the most likely failures are kept with the code and updated after every incident.
- Contact details of the recovery team, providers and partners are kept up to date.
- No single person holds the only access to a production system or secret.
7. Review
This policy is reviewed at least once a year and after every disaster recovery event or significant change to the architecture.