Recommended by RDEM Systems
MariaDB Replication with automatic failover — Signal18 Replication Manager
Every MariaDB cluster we operate relies on Replication Manager, the open source SQL replication software published by Signal18. It drives MariaDB failover, monitors replication and orchestrates daily backups.
The principle: automatic switchover driven by an external arbiter
In a MariaDB replication architecture, the problem isn't SQL replication itself — it's who decides the automatic switchover when the primary goes down.
Replication Manager solves this with a third-party monitoring instance, deployed outside the cluster. This instance continuously observes node status via their GTID (Global Transaction ID), detects failures, and triggers MariaDB failover in an automatic, controlled and traceable manner.
No handcrafted scripts, no human decisions under pressure at 3 AM. The external arbiter eliminates the risk of split-brain: only one node is promoted to primary, the others follow. Semi-synchronous replication ensures no committed transaction is lost during switchover.
Typical architecture
Primary node
Datacenter A
Reads + writes
Replica node
Datacenter B
Reads + failover
Replication Manager
Third-party instance
Monitoring + arbitration
What Replication Manager brings to your MariaDB cluster
MariaDB failover, MariaDB backup, SQL replication monitoring: the three pillars of a reliable cluster, automated by a single tool.
Automated failover
When the primary server fails, Replication Manager detects the failure and automatically switches to the replica. No human intervention, no time wasted at 3 AM.
Third-party monitoring instance
The failover is managed by an independent instance, outside the cluster. It observes, arbitrates and triggers the switchover — without split-brain conflict.
Daily SQL backup
A complete SQL dump is performed daily, validated and stored outside the cluster. Standard format, restorable anywhere — no proprietary format.
MariaDB replication monitoring
Continuous monitoring of replication lag, thread status and data consistency between nodes. Real-time alerts on each cluster.
Automatic rejoin
After a failure, the failed node is automatically resynchronized and reintegrated into the cluster. The topology rebuilds without manual intervention.
Open source & auditable
Replication Manager is free software. No black box, no proprietary license. The code is public, auditable, and maintained by Signal18.
Daily MariaDB backup in SQL format: your built-in disaster recovery plan
Every night, Replication Manager orchestrates a complete SQL dump of your MariaDB database. This backup is performed on the replica to avoid impacting primary performance.
SQL format is the universal standard for SQL replication: it restores on any MariaDB or MySQL instance, regardless of the target infrastructure. No proprietary binary format, no dependency on a specific tool.
Dumps are stored outside the cluster, on an independent backup infrastructure. In case of total disaster, you can rebuild your database from the latest backup — it's your built-in DRP (Disaster Recovery Plan), with a 24h RPO and an RTO of a few minutes.
With or without Replication Manager
Without Replication Manager
- ×Manual failover: human intervention required, unpredictable downtime
- ×Risk of split-brain without external arbiter
- ×Manual backups or fragile scripts, without automatic validation
- ×No real-time visibility on replication lag
- ×Manual resynchronization after failure — long and risky
With Replication Manager
- ✓Automatic failover in seconds, 24/7
- ✓Arbitration by third-party instance, reliable and traceable decision
- ✓Automated daily SQL dump, verified, stored outside the cluster
- ✓Real-time dashboard with alerts on lag and anomalies
- ✓Automatic rejoin: the cluster repairs itself
Why RDEM Systems recommends Replication Manager
We evaluated the alternatives — MHA, Orchestrator, MaxScale failover, in-house solutions. Replication Manager is the only tool that combines automated failover, integrated backup and MariaDB replication monitoring in a coherent, open source, and actively maintained solution. It natively interfaces with ProxySQL and HAProxy to route SQL traffic to the correct node after each switchover.
Signal18, the publisher, was founded by Stéphane Varoqui and Guillaume Lefranc, MySQL and MariaDB veterans with over 20 years of experience in MySQL replication, MariaDB replication and high availability.
In practice, RDEM Systems hosts and operates your cluster daily. For complex issues, we escalate directly to the Signal18 team. You benefit from two levels of expertise: operational and vendor.
Reliable, monitored, backed-up MariaDB replication
Let's discuss your current architecture and how Replication Manager can secure your database — without service interruption.
Start your managed MariaDB project
Let's discuss your database needs. Our DBA team advises you on the optimal architecture for your use case.
RDEM Systems SAS — SIREN 820 338 671 — 5 B rue des Noyers, 95300 Pontoise