Why it matters
Auto-inflate raises throughput units when traffic grows but never lowers them again, so namespaces often stay at their highest count long after a spike.How nOps finds it
nOps reads the last 14 days of Event Hubs billing. A namespace is flagged when it was billed for the same whole number of TUs on all 14 days and is not mostly covered by reservations. The target is the busiest day’s ingress events and operations, treated as if each were a 1 MB event, averaged over the day and rounded up to whole TUs (at least 1). The saving is the 7-day TU cost, scaled to a month, times the share of TUs removed. Services mostly paid with credits and savings under $5 a month are left out.Before you start
Access to the namespace’s metrics (incoming and outgoing bytes, throttled requests) to confirm the target.How to implement
- In the Azure portal, check the namespace’s Metrics for incoming and outgoing bytes and throttled requests at peak times, and how many consumer groups read the stream.
- Go to Scale, set the throughput units to the target shown, keep Auto-inflate on, and set its maximum to the current count so bursts are still absorbed.
- Or use
az eventhubs namespace update --resource-group RESOURCE_GROUP --name NAMESPACE --capacity TARGET_TUS --enable-auto-inflate true --maximum-throughput-units CURRENT_TUS. - Update Terraform, Bicep or ARM templates (
capacity) so a later deployment does not change it back.
Risks and considerations
- Daily billing shows neither data size nor peaks within a day. Check the metrics: short bursts above the target would be throttled without auto-inflate.
- Each TU also limits outgoing traffic (2 MB per second), so many consumer groups can need more TUs than ingress suggests.
- To roll back, raise the throughput units again; the change is immediate.
FAQs
Doesn't auto-inflate lower TUs by itself?
Doesn't auto-inflate lower TUs by itself?
No. Auto-inflate only scales up. You have to lower the count yourself.