---
title: "Deploy for Out-of-Band (OOB) Communications"
---
import useBaseUrl from '@docusaurus/useBaseUrl';
When a critical incident disrupts an organization’s primary infrastructure—due to a cyberattack, cloud outage, or internal failure—a secure, independent communication channel becomes essential. Mattermost can serve as that channel by operating in an Out-of-Band (OOB) environment: isolated, pre-provisioned, and ready to support incident response and business continuity teams.
This document outlines architectural guidance for deploying Mattermost in an OOB scenario across [generic](#generic-oob-architecture), [AWS](#aws-based-oob-deployment), and [Azure](#azure-based-oob-deployment) environments.
## Generic OOB architecture
This platform-agnostic design establishes a fallback communication system, decoupled from primary networks and identity providers.
### Architecture components
- **Users (Cybersecurity, IR Teams)**: Pre-authorized personnel trained to access and use the OOB platform during emergencies. Credentials and access methods should be provisioned in advance.
- **Secure Access Layer**: A firewall or access gateway protecting entry into the OOB environment. May include network policies, IP allowlists, or WAFs.
- **Secondary IdP / Emergency Auth**: A dedicated identity provider not dependent on the organization’s primary SSO or IdP (e.g., a minimal Azure AD tenant, Okta org, or locally stored credentials with MFA).
- **VPN / Bastion Host**: A hardened jumpbox or independent VPN granting access to internal OOB components and Kubernetes control plane.
- **Failover DNS Manager**: Services like Cloudflare Load Balancer, Azure Traffic Manager, or Route 53 to monitor primary system health and reroute DNS to the OOB ingress.
- **Ingress Controller**: e.g., ingress-nginx to handle HTTP(S) routing within the cluster, directing traffic to Mattermost Web, API, and WebSocket endpoints.
- **Mattermost Operator & App**: Automates lifecycle of Mattermost instances in Kubernetes; ensures updates, scaling, and configuration consistency.
- **PostgreSQL Database**: Persistent storage for channels, messages, users, and configurations; can be self-hosted or managed externally with hardened access.
- **Backups**:
- **Velero**: Backs up Kubernetes resources to external object store (S3, Blob).
- **pg_dump / Cloud-native backup**: Regular logical exports or managed backups.
- **Object Storage**: Durable, versioned storage (e.g., S3, Azure Blob) for snapshots and dumps.
## AWS-based OOB deployment
Deploying Mattermost in AWS leverages managed services and strong isolation capabilities.
### AWS components
- **Amazon EKS**: Managed Kubernetes for Mattermost Operator and application, with HA and auto-scaling.
- **Route 53**: DNS service for health-checked failover to the OOB environment.
- **Amazon Aurora PostgreSQL**: High-performance managed PostgreSQL with automated failovers.
- **Amazon S3**: Object storage for Velero snapshots, configurations, and database backups (with versioning and cross-region replication).
- **IAM Roles**: Scoped roles for EKS nodes and Velero using IRSA, enforcing least-privilege.
### Azure components
- **Azure Kubernetes Service (AKS)**: Runs Mattermost workloads with native autoscaling and Azure AD integration.
- **Azure DNS + Traffic Manager**: Global traffic routing and DNS failover based on endpoint health.
- **Azure Database for PostgreSQL**: Managed DB service with automated backups and patching.
- **Azure Blob Storage**: Destination for Velero snapshots and DB dumps; supports lifecycle policies and secure access.
- **Azure AD (Secondary Tenant)**: Acts as the IdP for OOB; deployed in a separate directory with scoped permissions and MFA.
| Domain | Recommendation |
|---|---|
| Access Control | Isolate authentication from primary infrastructure. Use MFA, secure vaults, and emergency tokens. Preload user accounts in the secondary IdP. |
| DNS Failover | Employ TTL-aware health checks via Route 53 or Traffic Manager. Monitor critical HTTP(S) or TCP endpoints. |
| Backup Strategy | Schedule frequent Velero and database backups. Store in versioned, encrypted object storage. |
| Disaster Recovery Drills | Conduct quarterly failover exercises: DNS switching, Mattermost access, and data restores. |
| Security | Treat OOB as Tier 0. Use hardened OS images, audit logging, strict RBAC, and centralized monitoring. Rotate secrets regularly. |