Project description
We run a busy web platform on a single database server, and that single point of failure keeps us awake at night. Any hardware fault, failed update, or heavy runaway query can take the whole product offline for thousands of customers at once. We are looking for a seasoned database engineer to design and build a fault-tolerant cluster that removes this risk and lets the team stop worrying about the next outage.
The core of the work is replication across at least two nodes with automatic failover, so that when the primary becomes unavailable, traffic switches to a healthy replica within a few seconds and without anyone waking up to do it by hand. We also need a solid, tested backup strategy: scheduled full and incremental copies, off-site storage, retention rules, and a documented restore procedure that we can actually run under pressure. Health checks and alerting must let us learn about trouble before our users do.
You should be comfortable with high-availability tooling, connection routing, and diagnosing replication lag. We expect a short architecture proposal first, then implementation on a staging copy, a controlled failover drill we watch together, and finally a careful production rollout. Documentation written in plain language, so our own engineers can operate the cluster, is a required part of the deliverable, not an afterthought.
The core of the work is replication across at least two nodes with automatic failover, so that when the primary becomes unavailable, traffic switches to a healthy replica within a few seconds and without anyone waking up to do it by hand. We also need a solid, tested backup strategy: scheduled full and incremental copies, off-site storage, retention rules, and a documented restore procedure that we can actually run under pressure. Health checks and alerting must let us learn about trouble before our users do.
You should be comfortable with high-availability tooling, connection routing, and diagnosing replication lag. We expect a short architecture proposal first, then implementation on a staging copy, a controlled failover drill we watch together, and finally a careful production rollout. Documentation written in plain language, so our own engineers can operate the cluster, is a required part of the deliverable, not an afterthought.
Administration Cash, Bank transfer
