This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(284.8, 69%, 37%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ETG%3C/text%3E%3C/svg%3E)
In order to understand costs, I created an Azure SQL database with auto pause delay of 1-hour and let it run with very low activity for a few days:
The cost per day showed around £2, which was to be expected.
However, our dev's complained that the start-up time was too slow when they hit the database for the first time after 1-hour, so I disabled auto-pause. The cost per day then increase to around £10.
The database however is configured as serverless which I thought meant that I wouldn't be charged hardly anything if there was no usage. Over the festive holiday, there was zero usage on the database, but it still cost nearly £10 per day. Switching auto-pause back on immediately dropped the cost back down (and brought back the dev's complaints!):
Can anyone please explain what causes the cost for a serverless database that has no user activity? Under low usage from our dev's the cost hardly rises, so this cost increase must be attributed to the auto-pause function being disabled. What resources are in use please outside user activity that create this cost? I have no jobs running or any other scheduled tasks.
I tried to run log analytics and other tools in the portal, with no data found in any of them.
Based on my understanding, the price is calculated based on the number of Vcores and the compute time used, which is measured in seconds.
As the database is not paused, it would consider the Vcore active for the entire duration and charge you accordingly.
When the database is paused, no compute charges are incurred
As I understand it, when the database is paused it is entirely evicted from the VM it runs on (or whatever it is), and is only stored on a disconnected disk somewhere. When you bring it alive, the disk is actually mounted on a machine. And that is the usage you are paying for, not the actual CPU time.
Or in another words: your expectations of what serverless meant were in error.
Thank you - that's makes sense. I was under the impression that vCores were only billed if in-use (e.g. actual DTU processing, rather than just sat idle but not auto-paused).
If I calculate costs based on vCores always being available, the maths equate to what I'm seeing IMHO means that your understanding is correct:
What I find interesting is that the pricing documentation states this...
Scenarios well suited for serverless compute
- Single databases with intermittent, unpredictable usage patterns interspersed with periods of inactivity, and lower average compute utilization over time.
...but actually its way more expensive to use serverless for a low-usage/intermittent database as opposed to a Standard S1 with, say, 20 DTUs.
I'd be interested to hear other peoples' experiences of this as I'm struggling to decide whether migrating 30 databases will give better cost benefits under SQLMI...
The fact that you're both MVP's and have a completely different interpretation screams to me that Microsoft are not doing a good job of explaining this!
If that page it does say:
. If the amount of CPU and memory used is less than the minimum amount provisioned for each resource, then the provisioned amount is billed.
Here you need to know that "provisioned" means how many vCores you have specified and not "how much you use". The latter interpretation is not close at hand, but "provisioned" is not a word that everyone may use.
Then it says under Minimum compute bill:
If a serverless database is not paused, then the minimum compute bill is no less than the amount of vCores based on maximum (minimum vCores, minimum memory GB * 1/3)
I would say that I find this unambiguous. But it is true that there is some technical lingo. And when someone originally explained Serverless to me, they used a more direct explanation. That is, the point is that you don't pay when the service is paused. And I already other Azure databases where I paid no matter if I used it or not.
But this is the charm with the cloud! You learn new things every day. And you may have to pay dearly for the lessons....