In Oracle Cloud Infrastructure (OCI), "rack-level segmentation" is the physical realization of Fault Domains. While a data center contains thousands of racks, OCI organizes them so that a single failure at the rack level—such as a power surge or a switch crash—cannot take down your entire application.

Here is how OCI segments its hardware at the rack level to ensure high availability.

1. The Anatomy of a Fault Domain

Every OCI Availability Domain (a physical data center) is divided into exactly three Fault Domains (FDs). A Fault Domain is essentially a "logical data center" within the physical one, consisting of distinct sets of racks.

• Independent Power: Each rack in a Fault Domain has its own Power Distribution Units (PDUs). If FD-1 loses power due to a rack-level electrical failure, FD-2 and FD-3 remain unaffected.

• Redundant Networking: Racks are segmented by their Top-of-Rack (ToR) switches. In OCI's architecture, resources in different Fault Domains do not share the same ToR switch.

• Physical Separation: Oracle ensures that hardware assigned to different FDs is physically distributed across the data center floor to prevent localized physical incidents (like a localized fire or cooling failure) from impacting more than one FD.

2. Anti-Affinity: The Core Purpose

The primary goal of rack-level segmentation is anti-affinity. This is the practice of ensuring that redundant components of your application are not placed on the same physical hardware.

• Compute Instances: When you launch multiple web servers, you should place them in different Fault Domains (FD1, FD2, FD3). OCI guarantees these servers will be on different physical racks.

• Database Clusters: For technologies like Oracle RAC, OCI automatically places the database nodes in different Fault Domains. This ensures that even if an entire rack fails, the database remains online.

3. Segmentation during Maintenance

Rack-level segmentation isn't just for "accidents"—it’s also for planned updates. Oracle follows a staggered maintenance policy:

• One FD at a time: Oracle never performs disruptive maintenance (like patching power controllers or network switches) on more than one Fault Domain in a region at the same time.

• Safety for Customers: If your workload is spread across all three FDs, you are guaranteed that at least two-thirds of your infrastructure will remain operational even during a major infrastructure update.

4. Hardware-Level Access Control (ToR ACLs)

Rack-level segmentation also has a security component. OCI implements Access Control Lists (ACLs) at the Top-of-Rack (ToR) switches.

• Port-Level Validation: The ToR switch knows which IP addresses are assigned to which physical ports.

• Spoofing Prevention: If a server in one rack tries to send traffic pretending to be an IP address from a different rack (segment), the physical switch drops the packet. This prevents "lateral threat proliferation" across the racks.

Summary of Rack-Level Benefits

Key Takeaway for Your Blog:

"Rack-level segmentation in OCI means your application isn't just 'in the cloud'—it's physically distributed. By leveraging Fault Domains, you ensure that a single failing power cable or a faulty network switch in one rack won't cause a total application outage."