When Your FHIR Capacity Plan Needs a Re-Baseline

A FHIR capacity plan that is never re-baselined slowly stops describing the deployment it was written for. Traffic mixes shift, retention policies change, partner integrations onboard. The plan that was defensible at year one drifts into fiction by year three. Re-baselining is the discipline that keeps the plan honest.

Naming the triggers in advance is what turns re-baselining from an ad-hoc project into a scheduled event. For related walkthroughs, the FHIR learning path collects the surrounding material.

The Sustained Divergence Trigger

The clearest signal is a projection versus observed divergence that persists. When the actual metric sits fifteen percent outside the projected bracket for two weeks running, the projection is wrong. That is not noise; that is a real drift in the underlying workload.

Deployments that track the divergence weekly catch the trigger before it becomes a problem. A pass through the site's FHIR node planner helps you rebracket the growth against the new observed numbers.

The New-Workflow Onboarding Trigger

Every new workflow that lands in production carries a capacity signature. Chart review workflows add read pressure. Ingestion workflows add write pressure. Analytics workflows add I/O pressure.

A plan built without the new workflow is missing a load line. Re-baselining should happen at every workflow onboarding regardless of the projection divergence. For the input list, capacity planning for a FHIR server: the inputs nobody documents walks through what to update.

The Partner Change Trigger

Partner integrations shift traffic mix. A new payer partner adds bulk exports. A new EHR partner adds ingestion. A retiring partner reduces one of both. Each partner change is a re-baseline trigger.

The disciplined pattern is to include partner-change events in the operational calendar and to require a plan update as part of the change. Plans that ship without the partner update drift from reality on the day the partner goes live.

The Retention Policy Trigger

Retention policy changes ripple through the capacity plan. Extending retention from three years to five doubles the storage line and adds a compound growth allowance. Shortening retention pushes storage down but often keeps the growth line unchanged because the archive was already sized for the longer horizon.

Every retention policy discussion should trigger a plan review. For the five-year framing that these policies interact with, growth curves for a FHIR archive over five years covers the shape.

The Staging Divergence Trigger

Staging environments that stop predicting production behavior are a re-baseline signal. When staging says the deployment can handle a load and production disagrees, the plan sitting behind staging is out of sync with reality.

For the specific practice of keeping staging aligned, sizing a staging environment that predicts prod behavior covers the operational side. Staging divergence usually means the traffic mix moved and the mix is missing from the plan.

The Scheduled Trigger

Even when none of the above fires, plans should re-baseline on a scheduled cadence. Quarterly for a light touch-up. Annually for a full rebuild. Skipping the scheduled review is how deployments end up managing yesterday's workload with today's failures.

The Public Change Log

Every re-baseline should be documented publicly. What triggered the review, what changed in the plan, what new numbers were committed. The change log becomes the artifact operations reads before an incident review.

Plans that live in a static document decay. Plans that carry a public change log stay legible. Owning the re-baseline cadence is what keeps the capacity conversation honest.

Op-art-moire diagram of a re-baseline trigger flowchart with a projection line, a divergence detection ring, a magenta trigger badge, and a downstream replan step feeding a new baseline node on a stark white background with dot-grid annotations

Sources