System configuration

Controller node, compute node, and storage node

• Controller node

  A node for accessing VSP One SDS Block for management operations. For details, see Installing and managing a controller node.

• Compute node

  A node that runs the application of the user and instructs input/output of user data to the storage node. For details, see Installing a compute node.

• Storage node

  • (Bare metal) A physical server that makes up VSP One SDS Block.

  • (Cloud) A virtual server that makes up VSP One SDS Block (EC2 instance).

    • (Cloud) Tiebreaker node.

      A storage node that has only the role as a cluster master node in Multi-AZ configuration. No storage controller is deployed. Also, this node has neither compute ports nor drives.

Network

• Control network

  • (Bare metal) Network that connects the controller nodes to storage nodes. Use this network to send management commands to VSP One SDS Block running on storage nodes and communicate with external services (such as SNMP managers and NTP servers). The storage nodes are connected to this network through their control ports. For the bare metal model, you can configure either redundant configuration or non-redundant configuration to connect the controller ports. To configure redundant configuration, enable the storage node teaming function utilizing multiple ports.

  • (Cloud) Network that connects the controller nodes to storage nodes. Use this network to send management commands to VSP One SDS Block running on storage nodes and communicate with external services (such as SNMP managers and NTP servers). The storage nodes are connected to this network through their control ports.

• Compute network

  Network that connects compute nodes to storage nodes. Use this network to send user data. The storage nodes are connected to this network through their compute ports.

  • (Bare metal) This network uses the iSCSI or NVMe/TCP protocol.

  • (Cloud) This network uses the iSCSI protocol.

• Internode network

  • (Bare metal) Network between storage nodes. Use this network to protect data by storing the same data in multiple storage nodes, exchange user data when capacity is balanced between storage nodes, and transmit management information between storage nodes. Configure a redundant configuration with multiple ports by using teaming on storage nodes.

  • (Cloud) Network between storage nodes. Use this network to protect data by storing the same data in multiple storage nodes, exchange user data when capacity is balanced between storage nodes, and transmit management information between storage nodes. The storage nodes are connected to this network through their internode ports.

• BMC network (Bare metal)

  Network that connects the storage node BMC and the controller node. Use this network to operate the BMC from the controller node. The storage nodes are connected to this network through their BMC ports.

  The BMC network must be able to communicate with the control network.

Roles of storage nodes

The storage nodes in the storage cluster are named depending on their assigned roles.

• Cluster master node (primary)

  This storage node supervises the entire storage cluster.

  (Bare metal) This node is assigned the storage cluster’s representative IP address.

  • The node has an interface for communicating with users and the management application.

  • The node centrally controls all the storage nodes in the storage cluster. Users and the management application can control each storage node by simply issuing commands to the cluster master node (primary).

• Cluster master node (secondary)

  This storage node takes over the cluster master node (primary) if the primary node fails.

• Cluster worker node

  Any storage node that is controlled by the cluster master node (primary).

Protection domain and fault domain

A protection domain is a unit that configures the data assurance format and internal processing I/O resource utilization. There is one storage pool in the protection domain, and one or more fault domains in the protection domain.

(Bare metal) A fault domain is a group of storage nodes sharing hardware such as a power line or network switch. By setting up multiple fault domains and isolating hardware such as power supply system and network switches for each fault domain, the storage system can continue to operate if another fault domain is successful, even if there is a hardware failure within one fault domain.

(Cloud) A fault domain is a group of storage nodes located in a single Availability Zone. In Single-AZ configuration, the number of fault domains is one, because the storage nodes comprising a cluster can only be located in a single Availability Zone. In Multi-AZ configuration, storage operation can continue even if the storage nodes in a fault domain collectively become abnormal, provided that another fault domain is normally operating.

However, even in a multiple fault domain configuration, if there is at least piece of hardware, such as a power supply system or network switch, that shares with other fault domains, the hardware failure can cause multiple fault domains to fail at the same time and cause operations to stop.

If hardware such as power supplies and network switches cannot be isolated for each fault domain, redundant configurations should be in place in case the non-isolated hardware fails.

The fault domain is set during the VSP One SDS Block setup.

(Bare metal)

Configuration file

VSP One SDS Block has the following configuration file:

Storage pools and volumes

A storage pool is a logical area for storing user data. Each pool is composed of multiple drives. You only need to manage the storage capacity of storage pools. You do not need to manage each drive or verify the physical boundaries of drives.

You can expand a storage pool by adding drives to an existing storage node or adding a storage node.

Volumes are logical devices to which user data is written or from which user data is read. Volumes are managed by the storage controller. When you write user data to a volume, it is written to the corresponding storage pool.

Capacity management of storage nodes by the storage controllers

Storage controllers are part of VSP One SDS Block processes that manage storage node capacities and volumes.

The capacity of each storage node is managed by the storage controllers. The storage controllers also monitor the system configuration and usage.

The total storage pool capacity is the sum of the individual capacities managed by the storage controllers.

As many storage controllers as storage nodes are provided. A storage controller is deployed across multiple storage nodes and is redundant to tolerate storage node failures (the figure shows an example of redundancy degree 3).

There are two redundancy degree settings for the storage controller: OneRedundantStorageNode (degree = 2) and TwoRedundantStorageNodes (degree = 3). One of these settings is automatically selected according to the setting of the user data protection method.

For details about management of storage controllers, see Managing storage controllers in the VSP One SDS Block Storage Administrator Guide.