System architecture

The basic building block of a StorageGRID Webscale system is the grid node. Nodes contain services, which are software modules that provide a set of capabilities to a grid node. The values and statuses for all the functions of the StorageGRID Webscale system are reported through attributes.

Grid nodes

The StorageGRID Webscale system uses four types of grid nodes:
  • Admin Nodes provide management services such as system configuration, monitoring, and logging. Each grid has one primary Admin Node and might have any number of additional Admin Nodes for redundancy.
  • Storage Nodes manage object data and metadata storage, including data protection.
  • API Gateway Nodes (optional) provide a load balancing interface to the StorageGRID Webscale system through which applications can connect to the system. Alternatively, a third-party HTTP load balancer can be used.
  • Archive Nodes (optional) provide an interface through which object data can be archived to an external archival storage system (tape or the cloud) for long-term storage.

This diagram shows how grid nodes are arranged within a StorageGRID Webscale system:


diagram described by surrounding text

Services

Each type of grid node provides specialized capabilities by running different software services.
Node Type Key Function Related Service
Admin Node Tracks system activity. Audit Management System (AMS).
Manages system-wide configuration. Primary Admin Node only. Configuration Management Node (CMN).
Powers the web-based Grid Manager administration interface. Network Management System (NMS).
Processes RESTful management API requests. Management Application Programming Interface (mgmt-api).
Collects and stores metrics. Prometheus.
Storage Node Maintains topology and authentication information. Administrative Domain Controller (ADC). Three per site.
Monitors object metadata storage. Distributed Data Store (DDS).
Processes object storage protocol requests and manages object data on disk . Local Distribution Router (LDR).
Integrates with external clouds to support Platform Services. Replicated State Machine (RSM). Three per site.
Federates user identities from LDAP and Active Directory. Identity Service (idnt). Three per site.
Stores and protects object metadata. Cassandra.
Manages tenant accounts. Account Service (acct). Three per site.
Integrates with OpenStack Keystone for authentication. Keystone Service (kstn). Three per site.
API Gateway Node Balances client connections to Storage Nodes. Connection Load Balancer (CLB).
Archive Node Communicates with external archival storage system (tape or the cloud). Archive (ARC).
All Node Types Monitors the operating system and underlying hardware. Server Status Monitor (SSM).

Attributes

Attributes report values and statuses for all of the functions of the StorageGRID Webscale system. Attributes and the values they report form the basis for monitoring the StorageGRID Webscale system.
Attributes for volumes on Nodes page

Attribute values

Attributes are reported on a best-effort basis and are approximately correct. Attribute updates can be lost under some circumstances, such as the crash of a service or the failure and rebuild of a grid node. For example, if an LDR service crashes after it has registered the ingest of a number of objects, but before it has committed the updated attribute value to disk, any attribute updates since the last commit are lost.

Propagation delays might slow the reporting of attributes within the StorageGRID Webscale system. Updated values for most attributes are sent to the StorageGRID Webscale system at fixed intervals. For this reason, it can take several minutes before an update is visible in the system, and two attributes that change more or less simultaneously can be reported at slightly different times.

Attribute values are available at the deployment, data center site, and grid node. Attributes at the deployment and data center site provide a convenient synopsis of information for multiple grid nodes.