[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"glossary-change-freeze::en":3,"gloss-cluster-change-freeze::en":23,"gloss-next-change-freeze::en":9},{"slug":4,"category":5,"name":6,"definition":7,"meta_desc":8,"faq":9,"schema_markup":9,"related":10},"change-freeze","cloud","Change Freeze","A change freeze is a declared period during which nothing is deployed to production except fixes for problems that are already happening. Companies call one around the dates where the cost of a bad hour is highest — a shopping peak, a tax deadline, a scheduled event the business has advertised, the days when half the staff are away and the other half are covering unfamiliar systems. The reasoning is sound: the chance of an outage rises with the rate of change, and the consequence of an outage rises with the calendar, so lowering the first while the second is high is a reasonable trade. It is also a trade with a bill attached, and the bill is worth naming. Work does not stop during a freeze; it accumulates. What ships the day the freeze lifts is a much larger batch than usual, assembled by people who have been queuing behind each other for weeks, and large batches are exactly the condition under which a deploy goes wrong and is hard to diagnose. A freeze therefore moves risk rather than removing it, and moves it to a moment that is calmer but not necessarily better staffed. Three things make the practice work rather than merely feel safe. Define the exception clearly, so a genuine fix does not need an argument at midnight, and record what was let through. Keep the freeze short and its boundaries specific, because a vague freeze quietly extends itself. And plan the unfreeze as an event of its own, with the backlog released in pieces rather than at once and someone watching each piece land, so that the accumulated batch is taken apart deliberately instead of arriving as a single change nobody can attribute. The version of this that quietly does harm is the freeze nobody declared: a team that stops deploying in December out of caution, without announcing it, gets every cost of a freeze and none of its benefits, because there is no exception rule, no end date and no plan for the batch that has been building up.","A change freeze halts deploys when a bad hour is most expensive. Why it moves risk to the unfreeze, and how to release the backlog in pieces.",null,[11,14,17,20],{"slug":12,"name":13},"canary-deployment","Canary Deployment",{"slug":15,"name":16},"error-budget","Error Budget",{"slug":18,"name":19},"incident-management","Incident Management",{"slug":21,"name":22},"runbook","Runbook",[24,28,32,35,38,41,45,48,51,54,57,60],{"slug":25,"category":5,"name":26,"updated_at":27},"air-gapped-deployment","Air-Gapped Deployment","2026-09-09T03:30:02+00:00",{"slug":29,"category":5,"name":30,"updated_at":31},"autoscaling","Autoscaling","2026-09-08T21:32:36+00:00",{"slug":33,"category":5,"name":34,"updated_at":31},"availability-zone","Availability Zone (AZ)",{"slug":36,"category":5,"name":37,"updated_at":31},"block-storage","Block Storage",{"slug":39,"category":5,"name":40,"updated_at":27},"bring-your-own-cloud","Bring Your Own Cloud (BYOC)",{"slug":42,"category":5,"name":43,"updated_at":44},"budget-alert","Budget Alert","2026-09-08T21:32:37+00:00",{"slug":46,"category":5,"name":47,"updated_at":27},"control-plane","Control Plane and Data Plane",{"slug":49,"category":5,"name":50,"updated_at":44},"cost-allocation-tag","Cost Allocation Tag",{"slug":52,"category":5,"name":53,"updated_at":44},"cost-chargeback","Cost Chargeback",{"slug":55,"category":5,"name":56,"updated_at":44},"cost-observability","Cost Observability",{"slug":58,"category":5,"name":59,"updated_at":31},"disaster-recovery","Disaster Recovery",{"slug":61,"category":5,"name":62,"updated_at":31},"edge-ai","Edge AI"]