Passer au contenu
Dernière mise à jour

Installation Requirements

The information provided in this document applies to enterprise-grade installations where service continuity and disaster recovery are mandatory requirements.

1. Cluster Architecture Overview

A SmartTSA installation is composed of two constituent appliances:

  • Main Site Appliance (HQ)
  • Disaster Recovery Appliance (DR)

These two appliances operate together as a single logical cluster.

The cluster is designed to:

  • Guarantee high availability
  • Ensure service continuity
  • Prevent downtime in case one appliance becomes unavailable
  • Provide redundancy at infrastructure level

Even though two physical appliances are deployed, from the client perspective the system behaves as a single service endpoint.

2. Network Architecture Requirements

SmartTSA clustering is based on a multicast communication model.

For the cluster to function correctly:

  • Both appliances MUST be physically connected to the same network switch
  • The network switch MUST support MAC multicast forwarding
  • Multicast traffic must not be filtered or blocked

⚠️ If multicast is not supported or incorrectly configured, cluster communication will fail.

3. High Availability Model

3.1 Single Cluster IP

All SmartTSA services belonging to a single installation form a cluster identified by:

  • A single Cluster IP address

Clients never connect directly to an individual appliance.

Instead, they connect through:

  • A DNS name associated with the Cluster IP

Example:

tsa.company.local → 192.168.1.100 (Cluster IP)

The DNS record must resolve to the cluster IP.

3.2 No External Load Balancer Required

SmartTSA does NOT require an external load balancer.

Incoming requests addressed to the Cluster IP are automatically forwarded to all nodes using:

  • MAC Address Multicast mechanism

How It Works

Each node:

  • Announces itself with its own Unicast MAC address
  • Shares a common Multicast MAC address used by the cluster

The switch:

  • Associates the Multicast MAC address with all switch ports where SmartTSA nodes are connected
  • Forwards packets destined to the Cluster IP to all cluster nodes simultaneously

This mechanism guarantees internal request distribution and HA behavior without additional components.

4. Mandatory High Availability Prerequisites

The following requirements MUST be satisfied:

  1. All SmartTSA appliances must belong to the same broadcast domain
  2. The switch must support MAC multicast
  3. A valid Multicast IP channel must be available
  4. N+1 IP addresses are required to form a cluster of N nodes

All IP addresses must:

  • Belong to the same network
  • Use the same netmask
  • Be routable inside the broadcast domain

5. Multi-Site Deployment (HQ + DR)

In a two-site configuration (HQ and DR), cluster-level HA is complemented by site-level failover.

Because the two sites are distinct, failover must be handled externally.

  1. Periodically test HQ availability (e.g., every 10 seconds) by sending a timestamp request
  2. If HQ becomes unreachable:
    • Switch traffic to DR
    • Use DNS failover rules OR
    • Apply appropriate network reconfiguration

This ensures:

  • Automatic disaster recovery
  • Minimal service interruption
  • Controlled failover logic