Google Cloud NetApp Volumes offers several service levels with predefined performance levels per provisioned GiB to meet a variety of application needs, from 16KiBps per GiB with Standard, all the way up to 128KiBps per GiB with Extreme. But what if you have a volume that needs even higher performance per GiB? This is where manual quality of service (QoS) can help you.
Manual QoS allows you to set volume throughput individually, within the throughput allocated by the provisioned storage pool capacity. This ability helps you better manage your cloud costs while delivering the performance your applications need. And you can change volume throughput instantly, without downtime or other impact to your clients.
Organizing your volumes to use manual QoS
As you might imagine, manual QoS doesn’t offer much for small storage pools, which don’t have much throughput to allocate to their volumes. But with larger pools, manual QoS starts to shine.
So start with a large pool and combine multiple volumes that can use the available throughput together:
- Small volumes with high throughput needs, like data for high-performance computing or database files. You might also have clone or cache volumes that need higher throughput than their size would automatically provide.
- Large volumes with low throughput needs like backups or archives.
It’s sort of like a game of Tetris – you fit together the volumes with large and small throughput requirements to use the available pool throughput. Or leave some unallocated space in the pool so that you have spare throughput that you can assign to volumes that need a quick performance boost. You can also use the pool service level to control how much throughput is available.
Using manual QoS
Manual QoS is supported with the Standard, Premium, and Extreme service levels. You can use the Google API, Google Cloud CLI, or Terraform to manage QoS in your storage pools and volumes. Google Cloud console is not currently supported.
Start by either creating a new manual QoS pool or converting an existing auto QoS pool to manual QoS. The conversion to manual is permanent, and each volume starts with the throughput it had automatically A manual QoS pool cannot be changed to an auto QoS pool. However, the volumes from a manual pool can be reassigned to an auto pool if they use the Premium or Extreme service level.
You then adjust individual volume throughput between 16KiBps and the maximum available throughput of the pool. Regular volumes support a maximum of 4.5GiBps, and large capacity volumes can go up to 30GiBps.
Here are some examples using gcloud. This example creates an 80TiB Premium manual QoS pool with available throughput of 5120MiBps.
gcloud netapp storage-pools create mybigpool --project=myproject --location=us-east4 --capacity=81920 --service-level=premium --qos-type=manual --network=name=myvpc
This example creates two volumes, a 5TiB volume with the maximum 4500MiBps, and a 70TiB volume with 400MiBps
gcloud netapp volumes create mysmallfastvol --project=myproject --location=us-east4 --storage-pool=mybigpool --capacity=5120 --protocols=nfsv3 –share-name=mysmallfastvol –throughput-mibps=4500
gcloud netapp volumes create mybigslowvol --project=myproject --location=us-east4 --storage-pool=mybigpool --capacity=71680 --protocols=nfsv3 –share-name=mybigslowvol –throughput-mibps=400
Put your disaster region to work!
Some customers use volume replication to another region to protect their data, but then the data just sits there waiting for a disaster. Your destination volumes may not need much performance while they are in a replication relationship. You can reduce their throughput to free up throughput that you can use to run workloads in that pool. If a disaster occurs, you can reallocate volume throughput or increase the destination pool size instantly to deliver production-level performance.
Manual QoS is generally available to all customers. Please give it a try and let us know what you think in the Comments below.
