Skip to content

First edition

Edition boundary: Community edition architecture overview and Enterprise edition only capability summaries are both present.

NAMRBD Platform Architecture

This HTML edition explains NAMRBD's public architecture for developers, operators, and reviewers. It starts from ordinary Linux block-device usage, then follows the request path through the kernel module, gateway, SBS cluster, metadata stores, logical storage substrate, backends, read views, Backup/DR control-plane state, security/compliance state, scoped Governance/WORM state, the optional iSCSI standard target gateway, and validation evidence.

NAMRBD expands to Network Attached Multipath Resilient Block Device and is pronounced [nae-mur-bee-dee]. The name is descriptive: network-attached access, multiple host/gateway paths, and resilient block-device behavior are first-order architecture concerns.

The name should also make these product features explicit:

  • network-connected block storage built as a scalable distributed cluster;
  • multipath connections for high availability between hosts and gateways;
  • standard iSCSI target access as a gateway option: Community includes the basic gateway, sbsctl iscsi, and LUN export capped at 3 distinct iSCSI-exported volumes, while iSCSI HA, MPIO/ALUA, advanced security/audit, and scale observability remain Enterprise-only;
  • scoped Governance/WORM support for block-native derived objects and userspace gateway sealed-target write rejection, without regulatory certification or object-store API compatibility claims;
  • automatic recovery as much as practical when internal components fail;
  • product direction that prioritizes Linux and Kubernetes (K8S) environments.

Kubernetes and CSI are covered as an integration case after the core Linux and storage model is established.

Reader path
Linux host path
-\>
Authority boundaries
-\>
Storage substrate
-\>
Read views and GC
-\>
Security/Compliance
-\>
Kubernetes case

Output Shape

The document is intentionally split into chapter-level HTML files. This keeps each chapter easy to review, edit, and replace while the index page remains the entry point.

Operations Guides

Standalone installation and operations guides for the HA metadata dependencies are also available in this HTML edition. They use the shared architecture stylesheet while remaining separate runbook-style documents.

Guide Operational Focus
NAMRBD installation guide Standard install, build, multi-node bring-up, gateway/SBS/kernel/CSI/iSCSI install flow, and install verification gates.
Developer build path Community source layout, command binary builds, repo-local test caches, and edition-boundary checks for GitHub contributors.
NAMRBD admin guide Component ownership, day-2 operations, observability, release guardrails, Backup/DR, performance, security, iSCSI, Governance/WORM evidence, troubleshooting, security notes, and iSCSI operations.
NAMRBD user manual Quick start, volume attach/use flow, snapshot/restore, enterprise Backup/DR, performance/security/Governance-WORM boundaries, iSCSI access, Kubernetes user flow, validation commands, and FAQ.
etcd HA cluster install and operations Gateway control-plane authority, Raft quorum, bootstrap, NAMRBD gateway and namrbdctl integration, backup, recovery, and member replacement.
TiKV HA cluster install and operations SBS metadata and optional RawKV object-store dependency, PD/TiKV quorum, TiUP deployment, sbs-service integration, maintenance, backup, and upgrade flow.

Primary Review Questions

Question Where To Start
Which component owns a state transition? Components And Ownership
How does a logical range become a physical payload object? Logical-To-Physical Mapping
Why are snapshots and clones safe after overwrite or discard? Read Views and Reachability And GC
Where does Backup/DR state live? Metadata Authority, Observability, and Edition Boundaries
Where does security/compliance state live? Metadata Authority, Observability, and Interface Specifications
Where does iSCSI fit without changing storage authority? Interface Specifications and Edition Boundaries
How does Kubernetes use NAMRBD without owning storage semantics? Kubernetes/CSI Integration Case
Where are component interface surfaces summarized? Interface Specifications
How does NAMRBD resolve tail latency under high concurrent load? Extreme I/O Performance Engineering