5.1 OCI Block Volume Service

5.1 OCI Block Volume Service

[!NOTE] Foundations Blueprint Focus: Storage is a core component of Domain 2 (Core OCI Services), accounting for substantial coverage on the OCI Foundations Associate (1Z0-1085-26) exam. Candidates are expected to master the functional distinctions between boot and block volumes, attachment modes, dynamic performance scaling through Volume Performance Units (VPUs), multi-attach clustering, volume groups, and the operational differences between backups and clones.

Enterprise cloud workloads require persistent, high-throughput, and low-latency storage that behaves identically to physical hard disk drives and solid-state storage arrays found in traditional enterprise on-premises data centers. In Oracle Cloud Infrastructure (OCI), this capability is provided by the OCI Block Volume service.

---

Fundamentals of OCI Block Volume

The OCI Block Volume service provides persistent, unformatted, raw storage blocks that can be dynamically mounted to compute instances (both Virtual Machines and Bare Metal servers). Unlike ephemeral instance storage that is destroyed when a server is decommissioned, Block Volumes exist independently of compute instance lifecycles, ensuring complete data persistence.

Key Architectural Characteristics

• Non-Volatile Persistence: Data stored on block volumes persists across instance reboots, stops, and terminations (when preservation settings are retained). You can detach a volume from an old instance and immediately attach it to a newly provisioned instance.
• NVMe Network Fabric: OCI block storage is connected to compute hosts across a flat, non-blocking, low-latency network fabric utilizing Non-Volatile Memory Express (NVMe) over Fabrics technology. This off-box architecture delivers consistent microsecond-level latency and eliminates storage I/O bottlenecks.
• Availability Domain Bound: Block volumes are created within a specific Availability Domain (AD). A block volume can only be attached to compute instances located in the exact same Availability Domain. Cross-AD mounting is not supported due to latency and physical fabric boundaries.
• Built-in Resilience and Durability: OCI automatically stores redundant copies of data across multiple physical storage servers within the Availability Domain, providing enterprise-grade durability and built-in protection against drive or rack failures.
• Encryption by Default: All block and boot volumes are encrypted at rest by default using industry-standard AES-256 encryption keys managed by Oracle, or optionally with customer-managed keys stored in the OCI Vault service. In-transit encryption across the network fabric is also enabled by default on supported compute shapes.

---

Boot Volumes vs. Block Volumes

OCI categorizes block-level storage into two distinct functional types:

+-------------------------------------------------------------------------+
|                        OCI Block Storage Architecture                   |
|                                                                         |
|  +-----------------------------+       +-----------------------------+  |
|  |         Boot Volume         |       |         Block Volume        |  |
|  +-----------------------------+       +-----------------------------+  |
|  | - Created with Compute      |       | - Provisioned Independently |  |
|  | - Holds OS, Kernel, Boot    |       | - Formatted by Guest OS     |  |
|  | - Size: 50 GB to 32 TB      |       | - Size: 50 GB to 32 TB      |  |
|  | - 1 Boot Volume per Instance|       | - Up to 32 Volumes/Instance |  |
|  +--------------+--------------+       +--------------+--------------+  |
|                 |                                     |                 |
|                 +------------------+------------------+                 |
|                                    v                                    |
|                   [ OCI Compute Instance (VM / BM) ]                    |
+-------------------------------------------------------------------------+

1. Boot Volumes

• Automatic Lifecycle: When you launch a compute instance, OCI automatically creates and attaches a Boot Volume to that instance.
• Contents: Holds the operating system, system kernels, boot partition, and base system configuration files.
• Instance Deletion Behavior: When terminating a compute instance, OCI presents an option to either terminate or preserve the boot volume. Preserving the boot volume allows you to retain all installed applications and configurations, attach it as a secondary data disk to another instance for troubleshooting, or launch a replacement instance directly from it later.
• Capacity Range: Sizes range from 50 GB up to 32 TB.

2. Block Volumes (Data Disks)

• Independent Lifecycle: Provisioned separately as standalone storage resources and attached to running compute instances as secondary data drives.
• Guest OS Management: Exposed to the operating system as an unformatted block device (e.g., /dev/sdb on Linux or Disk 1 on Windows). The system administrator creates partitions, formats the disk with a chosen file system (e.g., XFS, ext4, NTFS), and mounts it to a directory tree.
• Capacity & Scaling: Each block volume can range in size from 50 GB up to 32 TB (in 1 GB increments). A single compute instance can have up to 32 block volumes attached simultaneously, providing up to 1 Petabyte (PB) of high-performance block storage per instance.

---

Attachment Types and Access Modes

When connecting a block volume to a compute instance, you must configure the attachment method and access permissions:

Attachment Protocols

1. iSCSI (Internet Small Computer System Interface):
   • Connects the volume to the instance over TCP/IP storage fabric.
   • Manual OS Setup Required: After attaching via the OCI Console or API, you must log in to the instance operating system and execute the specific iscsiadm configuration commands generated by OCI to discover and log in to the iSCSI target portal.
   • Performance: Delivers maximum raw I/O throughput and IOPS; mandatory for Bare Metal compute instances.
2. Paravirtualized:
   • Leverages hypervisor-emulated block drivers directly exposed to guest virtual machines.
   • Plug-and-Play Simplicity: Requires zero commands inside the operating system. The volume appears automatically as a block device upon attachment.
   • Ideal for virtual machines where streamlined operational simplicity is preferred over maximum edge I/O performance.

Access Modes & Multi-Attach Capabilities

• Read/Write (Default): Grants full exclusive read and write permissions to the instance.
• Read-Only: Restricts the instance to read-only operations, preventing accidental data modification.
• Multi-Attach (Shared Block Storage):
  • OCI allows a single block volume to be attached to up to 8 compute instances concurrently in either Read/Write or Read-Only mode.
  • Clustered Workloads: Multi-attach is specifically engineered for shared-disk clustering architectures, such as Oracle Real Application Clusters (RAC) and clustered file systems (e.g., OCFS2, GFS2).

[!IMPORTANT] Exam Trap Alert: Attaching a single block volume to multiple instances in Read/Write mode without a cluster-aware file system (such as OCFS2) will result in filesystem corruption because independent operating system kernels will overwrite each other's block buffers without coordination.

Volume Performance Units (VPUs) & Dynamic Performance Scaling

In traditional cloud environments, increasing disk performance requires resizing the disk to an unnecessarily massive capacity or migrating data to a completely different storage family. OCI eliminates this friction through Volume Performance Units (VPUs).

How VPUs Work

Performance in OCI Block Volume is decoupled from disk capacity. Administrators configure performance on a per-gigabyte basis by allocating Volume Performance Units (VPUs/GB). OCI provides four distinct performance tiers:

1. Low Cost (0 VPUs/GB):
   • Performance: Provides 2 IOPS per GB up to a maximum of 3,000 IOPS per volume, and 240 KB/s per GB throughput up to 480 MB/s.
   • Target Workloads: Cost-optimized tier designed for throughput-intensive sequential workloads with large I/O sizes, such as streaming media, log data ingestion, backup targets, and batch processing pipelines.
2. Balanced (10 VPUs/GB - Default):
   • Performance: Delivers 60 IOPS per GB up to 25,000 IOPS per volume, and 480 KB/s per GB throughput up to 480 MB/s.
   • Target Workloads: The standard default tier offering balanced price-to-performance for web servers, enterprise application middleware, e-commerce platforms, and general-purpose business systems.
3. Higher Performance (20 VPUs/GB):
   • Performance: Delivers 75 IOPS per GB up to 50,000 IOPS per volume, and 600 KB/s per GB throughput up to 480 MB/s.
   • Target Workloads: High-I/O transactional databases (Oracle Database, Microsoft SQL Server, PostgreSQL) and mission-critical enterprise ERP systems.
4. Ultra High Performance (30 to 120 VPUs/GB):
   • Performance: Configurable in 10 VPU increments from 30 up to 120 VPUs/GB, scaling performance up to an incredible 300,000 IOPS and 2,680 MB/s throughput per volume.
   • Target Workloads: Extreme performance-intensive enterprise workloads, financial algorithmic trading engines, massive analytics databases, and Exadata-adjacent transactional systems.

OCI Block Volume Performance Tier Comparison

[!TIP] Critical Exam Concept: Volume Performance Units can be dynamically adjusted up or down in seconds via the OCI Console, CLI, or REST API without detaching the volume, shutting down the database, or rebooting the compute instance! For example, an e-commerce platform can dynamically scale block volumes to Higher Performance on Black Friday and scale back to Balanced on Monday with zero downtime.

---

Volume Groups

In complex multi-tier architectures, applications frequently distribute data across multiple block volumes and compute instances (for example, separate volumes for database data files, redo logs, index tablespaces, and application binaries). Taking individual snapshots of these volumes sequentially risks inconsistent recovery states.

A Volume Group allows administrators to group multiple boot and block volumes together into a single manageable entity:

+-------------------------------------------------------------------------+
|                        OCI Volume Group Architecture                    |
|                                                                         |
|  +-------------------------------------------------------------------+  |
|  |                        Production Volume Group                    |  |
|  |  [Boot Vol 1]  [DB Data Vol]  [DB Redo Log Vol]  [App Binaries]   |  |
|  +-----------------------------------+-------------------------------+  |
|                                      |                                  |
|              Coordinated Atomic Snapshot / Clone Action                 |
|                                      v                                  |
|  +-------------------------------------------------------------------+  |
|  |            Coordinated Crash-Consistent Point-in-Time             |  |
|  |                Backup or Clone Across All Volumes                 |  |
|  +-------------------------------------------------------------------+  |
+-------------------------------------------------------------------------+

Key Capabilities of Volume Groups

• Crash-Consistent Point-in-Time Snapshots: OCI coordinates an atomic snapshot across all volumes in the group at the exact same point in time, guaranteeing crash-consistent state across multi-disk databases.
• Coordinated Cloning: Clone an entire multi-volume application or multi-tier environment into a staging or development environment with a single operation.
• Simplified Disaster Recovery: Volume groups can be backed up to another OCI region on an automated schedule, streamlining business continuity planning.

---

Data Protection: Backups vs. Clones

OCI Block Volume provides two complementary data protection and replication mechanisms:

1. Block Volume Backups

• Target Storage: Block volume backups are stored encrypted and durably in OCI Object Storage, completely decoupled from the underlying block volume storage infrastructure.
• Backup Types:
  • Full Backup: A complete, standalone copy of all data blocks on the volume.
  • Incremental Backup: Captures only the data blocks that have changed since the previous backup. Faster to create and consumes substantially less storage.
• Cross-Region Replication: Backups can be copied asynchronously to another OCI region to support geographic disaster recovery.
• Policy-Based Automated Backups: Instead of scheduling manual cron scripts, administrators assign predefined or custom Backup Policies to volumes or volume groups:
  • Bronze Policy: Monthly full backup retained for 12 months.
  • Silver Policy: Weekly full backup (retained 4 weeks) + daily incremental backup (retained 30 days).
  • Gold Policy: Daily incremental backup (retained 7 days) + weekly full backup (retained 4 weeks) + monthly full backup (retained 12 months) + annual full backup (retained 5 years).

2. Block Volume Clones

• Target Storage: Clones are created directly on the Block Volume storage hardware within the same Availability Domain.
• Instant Provisioning (Copy-on-Write): A clone is a point-in-time direct volume-to-volume duplicate. Leveraging copy-on-write technology, the clone becomes accessible and mountable in seconds, regardless of whether the source volume is 100 GB or 32 TB.
• Zero Impact on Production: Creating a clone does not pause or degrade the active I/O performance of the source volume.
• Common Use Cases: Rapidly provisioning ephemeral staging or testing environments, spinning up development sandbox databases, and executing destructive patch testing without risking production data.