1.2 OCI Global Architecture: Regions, Availability Domains & Fault Domains

1.2 OCI Global Architecture: Regions, Availability Domains & Fault Domains

[!NOTE] Architectural Hierarchy: Oracle Cloud Infrastructure organizes physical and logical assets into four distinct tiers: Realm -> Region -> Availability Domain (AD) -> Fault Domain (FD). Every resource provisioned in OCI exists within this hierarchy, determining its physical placement, fault isolation, and disaster recovery scope.

Oracle Cloud Infrastructure is engineered from the ground up to support high-performance enterprise workloads, database clustering, and mission-critical applications. To eliminate systemic failures and satisfy diverse regulatory and compliance mandates worldwide, OCI structures its physical infrastructure across four distinct hierarchical tiers.

+-------------------------------------------------------------------------------+
|                                    REALM                                      |
|  (Completely isolated: OC1 Commercial, OC2/OC3 Government, OC20 EU Sovereign) |
+-------------------------------------------------------------------------------+
                                        │
                                        ▼
+-------------------------------------------------------------------------------+
|                                   REGION                                      |
|    Localized geographic area (e.g., US East Ashburn, Germany Central Frankfurt)|
+-------------------------------------------------------------------------------+
                                        │
                                        ▼
+-------------------------------------------------------------------------------+
|                           AVAILABILITY DOMAIN (AD)                            |
|    Fault-isolated physical datacenters connected by ultra-low-latency fiber   |
+-------------------------------------------------------------------------------+
                                        │
                                        ▼
+-------------------------------------------------------------------------------+
|                           FAULT DOMAINS (FD-1, FD-2, FD-3)                    |
|    Grouping of hardware (servers, ToR switches, PDUs) inside an AD (Anti-Affinity)|
+-------------------------------------------------------------------------------+

---

OCI Realms

A Realm is a logical collection of regions that are completely isolated from all other realms. Realms share zero customer data, zero encryption keys, zero identity principals, and zero network routing paths. This total separation guarantees that a security incident, identity compromise, or administrative configuration in one realm cannot propagate to another.

OCI operates several distinct realms:

1. Commercial Realm (OC1): The primary global realm hosting commercial enterprises, startups, and academic institutions worldwide (oraclecloud.com).
2. Government Realms (OC2, OC3, OC4): Purpose-built realms designed to meet stringent federal regulatory standards. For example, the US Federal Government realm (OC2) complies with FedRAMP High and DISA Impact Level 5 (IL5), while the US Department of Defense realm (OC3) complies with DISA Impact Level 6 (IL6) for classified secret workloads.
3. Sovereign Realms (e.g., OC20): Dedicated sovereign cloud realms, such as the Oracle EU Sovereign Cloud, architected to meet rigorous European Union privacy and data localization laws (such as GDPR). These realms are operated exclusively by EU-based legal entities employing EU citizens.

---

OCI Regions

An OCI Region is a localized geographic area containing one or more Availability Domains. Examples include US East (Ashburn), US West (Phoenix), Europe Central (Frankfurt), and Asia Pacific (Tokyo).

Single-AD vs. Multi-AD Regions

• Multi-AD Regions: OCI's foundational regions (such as Ashburn, Phoenix, Frankfurt, and London) contain three physical Availability Domains (AD-1, AD-2, AD-3) separated by miles to withstand localized environmental disasters while maintaining ultra-low-latency interconnectivity.
• Single-AD Regions: Most newer OCI regions are designed as high-density single-AD regions. In a single-AD region, high availability is achieved locally using the three internal Fault Domains, and disaster recovery is achieved by pairing with another nearby regional datacenter.

Region Subscription

A customer tenancy has a primary home region where identity resources and tenancy subscriptions originate. To deploy resources in additional regions, an administrator must explicitly subscribe the tenancy to those target regions. Once subscribed, network resources, compute instances, and storage can be deployed in the newly accessible regions immediately.

Data Sovereignty and Data Residency

A core guarantee of OCI is data sovereignty:

• Customer data stored in an OCI region (such as Block Volumes, Object Storage buckets, or Database tables) never leaves that region unless the customer explicitly configures cross-region replication, backup policies, or egress networking.
• Oracle does not automatically move, replicate, or inspect customer data across national borders or regional boundaries. This ensures compliance with regional data residency legislation (such as HIPAA in the United States or GDPR in Europe).

---

Availability Domains (ADs)

An Availability Domain (AD) consists of one or more discrete, physical data centers located within a geographic region.

Fault Isolation

Availability Domains are completely isolated from one another in physical infrastructure:

• Independent Utilities: Each AD possesses independent municipal power feeds, backup generator plants, cooling chillers, and physical security perimeters.
• Resilience: A flood, localized power grid failure, or physical catastrophic event at one AD does not affect the operation of neighboring ADs within the same region.
• No Shared Failure Modes: ADs share no single points of physical failure.

Ultra-Low-Latency Backbone Network

Despite physical isolation, all Availability Domains within a region are interconnected by an Oracle-engineered, private, high-speed optical network backbone.

• Sub-Millisecond Latency: Round-trip network latency between ADs within the same region is typically less than 1 millisecond.
• Synchronous Replication: This ultra-high throughput and negligible latency allows enterprises to run synchronous active-active database clustering (such as Oracle Real Application Clusters / RAC) and synchronous storage replication across multiple ADs without application performance degradation.

---

Fault Domains (FDs)

While Availability Domains protect against catastrophic data center outages, physical components within a single data center can still fail. To provide granular hardware resilience, OCI introduces Fault Domains.

A Fault Domain (FD) is a grouping of physical hardware and infrastructure within an Availability Domain.

The Rule of Three

Crucial exam rule: Every Availability Domain contains exactly three (3) Fault Domains, designated as:

• FAULT-DOMAIN-1
• FAULT-DOMAIN-2
• FAULT-DOMAIN-3

In a single-AD region, there are 3 Fault Domains. In a multi-AD region with 3 ADs, there are a total of 9 Fault Domains (3 per AD).

Anti-Affinity and Component Isolation

Fault Domains enforce physical anti-affinity. Compute instances placed in different Fault Domains are guaranteed not to share:

1. Physical server hardware (different motherboards, CPUs, and hypervisor hosts).
2. Top-of-Rack (ToR) network switches.
3. Power Distribution Units (PDUs) and power rails.

Dual Protection: Hardware Failures and Planned Maintenance

Fault Domains protect enterprise workloads in two vital ways:

1. Unexpected Hardware Outages: If a physical server experiences a power supply blowout, RAM ECC failure, or a ToR switch crashes in FAULT-DOMAIN-1, instances running in FAULT-DOMAIN-2 and FAULT-DOMAIN-3 continue operating normally.
2. Planned Maintenance: Oracle frequently updates hypervisor kernels, server firmware, and network switch software. Oracle orchestrates rolling maintenance one Fault Domain at a time. While FAULT-DOMAIN-1 undergoes maintenance, FAULT-DOMAIN-2 and FAULT-DOMAIN-3 remain fully operational.

When provisioning a Compute instance, you can either select a specific Fault Domain or allow OCI to automatically distribute instances across available Fault Domains.

---

Architectural Comparison: Region vs. AD vs. FD

---

Design Patterns for High Resilience

To build resilient enterprise architectures on OCI, architects combine Regions, ADs, and FDs into structured defense-in-depth patterns:

Pattern 1: High Availability Within an AD (Single-AD or Multi-AD)

Deploy application tier instances across all three Fault Domains (FD-1, FD-2, FD-3). Place an OCI Load Balancer in front of the instances to distribute web traffic and automatically route around an unhealthy instance if an individual Fault Domain experiences an outage or maintenance event.

Pattern 2: High Availability Across ADs (Multi-AD Regions)

Deploy active application servers in AD-1, AD-2, and AD-3. Configure a public OCI Load Balancer, which automatically provisions redundant listeners across multiple ADs. For the database tier, deploy an active-active or active-passive cluster (such as Oracle Autonomous Database with Autonomous Data Guard or Oracle RAC) across different ADs to achieve zero-downtime failover.

Pattern 3: Cross-Region Disaster Recovery

For catastrophic business continuity, configure cross-region asynchronous replication between geographically distant regions (for example, Ashburn and Phoenix). In the secondary region, maintain a "pilot light" or warm standby environment ready to assume production traffic using OCI Traffic Management Steering policies or DNS failover if the primary region suffers a complete outage.