Scaleway Instances now integrate placement groups without any extra cost for Development, General Purpose, ARM and GPU ranges. This power feature lets you decide where to place your virtual instances.

Previously, when you created new instances at Scaleway, our allocator placed them without taking into account your existing resources. This could be restrictive in some use cases as it happens that specific needs require particular placements: on the same hypervisor or, on the contrary, on different ones.

As well as facing this problem for our internal needs, several customers reported this issue. Setting-up high availability on our services Database and Kubernetes Kapsule, we indeed had to make sure that the provisioned instances were not on the same physical server. Those reasons led us to the release of this feature today

Placement groups offer two possibilities:

  • Low-latency groups creation: your instances are located as close as possible so as to leverage network throughput at its best.
  • High-availability groups creation: your instances are placed on distinct physical hypervisors so as to maximise availability of your application.

In addition, the policy of your placement groups can be enforced or optional:

  • If you choose to enforce placement, your policy will be strictly applied. However, your instances may not start if our allocator is not able to respect your choice.
  • If you choose to make placement optional, your instances will start no matter the group policy can be respected or not. Our allocator will do its best so that your instances meet the group criteria and let you know when it’s not the case.

To create or modify a placement group, your instances must be powered off. They may be moved at the next start to ensure that your placement policy is respected.

You can create and manage your placement groups directly from the console or using our API. Here are both documentations to guide you:

Happy cloud riding!


A Few Words About Scaleway Instances

Scaleway Instances are virtual cloud instances delivered in seconds with backup, network and security options. They offer up to 48 vCPUs and 256 GB of RAM at the best price-performance ratio to let you create, deploy and scale your infrastructure in the cloud. Learn more…