Reputation: 365
I can't seem to find concrete rate limits for cloning disks in GCP. From the docs I see:
But what is the limit for different disks? i.e. I want to clone 500 different disks one time. Assuming this doesn't go over the 1TB per 15 minutes, how many total clone requests can I make.
Also if I do a burst request of 257TB when does that limit come back? Is it 15 minutes like the 1TB per 15 minutes limit?
Upvotes: 0
Views: 229
Reputation: 3281
You can keep track of the API rate limits in Stackdriver monitoring as there is a metric for API usage available, they can create a request count and filter it by service.
You can clone at most 257TiB for a particular project and zone. After that, your quota is replenished @ 1TiB/15min. So you will be able to work around this 1 TiB/15min clone disk space limit too if you delete the disk right after creating the clone.The limit is for how many existing devices can have been created via cloning. If you aren't using the clones except to stress drainwrites/agents, then you can just constantly delete any cloned devices
The public doc states that the cloning throughput limits only apply to zonal to regional (Z2R) clones
. Zonal to zonal and regional to regional only have time based rate limits and a max of 1000 clones per device.
For total clone requests that depends on the disk type and your project's resource quotas and the maximum sustained disk throughput for a standard persistent disk.To get an exact count of the total clone requests raise a support ticket so that they can calculate based on your project and resources.
Snapshotting large amounts of persistent disk might take longer than expected to complete and might provide an inconsistent view of your logical volume without careful coordination with your application. As mentioned in this doc as a best practice, take a snapshot of the disk once per hour
. Avoid taking snapshots more often than that. FYI: You can have a total attached capacity of 257 TB per instance
.
Upvotes: 1