We are excited to introduce fault domains, a new way to manage and improve availability for Oracle Cloud Infrastructure Virtual Machine and Bare Metal compute instances within an Availability Domain.
Today you can use Availability Domains to help ensure high availability for your applications, by distributing virtual machine (VM) and bare metal instances across multiple availability domains within a single region. Availability Domains are physically isolated and do not share resources (power, cooling, network), which means the likelihood of multiple availability domains within a region failing is very small. The use of multiple Availability Domains ensures high availability because a failure in any one availability domain won't impact the resources running in the others.
If you want more granular control of application availability within a single Availability Domain, you can now achieve that by using fault domains. Fault domains enable you to distribute your compute instances so that they are not on the same physical hardware within a single Availability Domain, thereby introducing another layer of fault tolerance. In addition to host anti-affinity, you get a level of power anti-affinity since the physical hardware has independent and redundant power supplies preventing power faults within a Fault Domain from affecting other Fault Domains.
Fault domains can protect your application against unexpected hardware failures or outages caused by maintenance on the underlying compute hardware. Additionally, you can launch instances of all shapes within a fault domain.
Oracle Cloud Infrastructure is typically designed with three availability domains per region, and each availability domain has three fault domains. When carrying out maintenance on the underlying compute hardware, Oracle Cloud Infrastructure ensures that only a single fault domain is impacted at one time to guarantee availability of your instances in the remaining fault domains.
Getting started is easy. When you create a new compute instance using the API, CLI or Console, you can specify the fault domain in which to place the instance. If you don’t specify a fault domain, the instance will be distributed automatically in one of the three fault domains within that availability domain. Oracle Cloud Infrastructure makes a best-effort anti-affinity placement while optimizing for available capacity in the Availability Domain.
To modify the fault domain after an instance has been created, you must terminate and re-create the instance. All existing VM and bare metal instances have been distributed automatically among the three fault domains in the their availability domain.
The instance details page shows the fault domain information along with other metadata about the instance.
To get started with fault domains on Oracle Cloud Infrastructure, visit https://cloud.oracle.com. Fault domains are available at no additional cost in all public regions. For more information, see the Oracle Cloud Infrastructure Getting Started guide, Compute service overview, Compute FAQ, and Fault Domains documentation for more information.