Multiple service instances per host

Intent

Run multiple service instances on the same host to improve utilization/density.

Use when

• Cost/utilization pressure matters and you can tolerate some shared-blast-radius risk.
• You have good isolation controls (resource limits, cgroups, QoS, noisy-neighbor protections).

Avoid / watch-outs

• Noisy-neighbor and shared resource contention are common failure modes.
• Debugging can get harder; ensure per-instance observability and clear resource metrics.

Skill mapping

• observability: host-level + service-level metrics and correlation for contention incidents.
• resilience: bulkheads/concurrency limits to avoid self-saturation under contention.
• platform: standardize deployment configs/limits so every service doesn’t reinvent them.