Redundancy starts before an outage
For an application to keep responding when a server disappears, healthy instances need to take over. They need sufficiently independent hosts, available resources and a functioning network entry point.
NodalX connects redundancy policy to instance placement and health checks. Machine count alone is not enough: three VMs on one host can fail together.
What happens when a host is lost
When an instance no longer passes checks, its route can be withdrawn. New requests reach healthy instances according to the project policy. Reserve capacity can then rebuild resources where possible.
A lost server remains lost until it is restored. Traffic redirection does not promise to replay every in-flight request or make every stateful application switch without interruption.
Data has its own continuity
A database, files and a job queue do not replicate in the same way. Each stateful service needs appropriate write, replication and promotion rules. A recovery policy needs to identify the available data point and how to verify the result.
Backups protect against different risks from redundancy. An application error may spread to all active copies; a separate backup can recover an earlier state.
Prepare planned stops too
Maintenance can use the same mechanisms: prepare replacement capacity, check its instances, move traffic and drain the old host. Reserve capacity serves routine fleet maintenance as well as incidents.
Protection levels still depend on topology and available resources. NodalX presents these operations in the project’s context so choices and limitations remain visible.