MicrosoftDocs
Unclear example of cached VM disk limits
In my opinion, the last two examples of cached disk accesses are unclear, or potentially misleading (depending whether I understood it correctly or not).
Look at this example:
The described setup is this: Standard_D8s_v3 Cached IOPS: 16,000 Uncached IOPS: 12,800
If the disk IO requests have a cache hit, then I understand that the total 3x5000 IOPS is below the cached limit of 16000 IOPS. But in this case, why are there arrows going from the cache to the disk? I think this would be incorrect, since the data comes only from cache, not from the attached disks.
But on the other hand, if the data indeed comes from the attached disks as the arrows would imply (i.e. cache miss), then according to the docs, a few paragraphs above the referenced image:
When a read is performed and the desired data is not available on the cache, the read request is relayed to the disk. Then the disk surfaces it to both the cache and the VM. This read is counted toward both the VM's uncached limit and the VM's cached limit.
In that case, I would assume, that the VM uncached limit of 12800 IOPS is reached and the maximum requested IOPS of 15000 is not achieved.
A third interpretation would be that even when reading from cache, somehow disk IOPS limits apply as well? But I can't see this documented explicitly like this anywhere.
Is there a mistake in my reasoning, or is the documentation incorrect here?
Document details
⚠ Do not edit this section. It is required for learn.microsoft.com ➟ GitHub issue linking.
- ID: 8d99436a-dbb2-a88a-9881-2472e4c5ca80
- Version Independent ID: 2a2b5781-fbd1-a8ee-0aac-5aed7540d57a
- Content: Virtual machine and disk performance - Azure Virtual Machines
- Content Source: articles/virtual-machines/disks-performance.md
- Service: storage
- Sub-service: disks
- GitHub Login: @roygara
- Microsoft Alias: rogarana
@jhyot Thanks for your feedback! We will investigate and update as appropriate.
@jhyot Thanks for bringing this to our attention. I'm going to assign this to the document author so they can take a look at it accordingly
@roygara Can you please check and add your comments on this doc update request as applicable.
@jhyot Thanks for reaching out and bringing this to our attention. We appreciate it!
In this case the article is incorrect, we've made some updates and they should be live in the next day. I'll let you know when they're live and close out this issue then.
Again, thanks for bringing this to our attention and helping us improve the docs! :)
@jhyot The updates are live. Accordingly, I'll close this out. Thanks again!
#please-close