This section introduces the HA service, including supported models, the architecture, system requirements, and limitations. Make sure you go through this section before you start using the HA service.

About HA Service

The HA service is a storage solution that uses two storage devices at different sites to ensure continuous service. In the HA architecture, identical copies of data are stored on both devices. If one device malfunctions, the other one immediately takes over the service to eliminate the risk of data loss or downtime. HA service thus makes a suitable choice for mission-critical workloads that require non-stop operations with a low tolerance for downtime.

Supported Models

The HA service is supported on the following storage series.

• GS family: 3000 Gen2, 4000 Gen2, 3000B Gen2, 4000B Gen2, 3000 G3, 4000 G3, 3000U, 3000UT, 4000U, 5000U
• GSe family: 3000 Gen2, 4000 Gen2, 3000 G3, 4000 G3, 3000U, 3000UT, 4000U

Architecture

The HA storage consists of the following major components that are connected via network links.

• Two storage devices (at site A and site B): The storage devices provide data services. When data is being written to one device, it is replicated to the other device at the same time, so that data is constantly synced between the two devices.
• Witness server (at site C): When a failover situation occurs, the witness server determines which device should take over services to ensure data availability on the HA storage.
• Hosts: The HA service solution is applicable to databases (Oracle and SQL) and virtual clusters (such as VMware).

*Auto-failback is available for block-level HA service only.

System Requirements

Hardware Requirements

• The two storage devices must be of the same model, with identical memory and host board configurations.
• Only SATA SSD, SAS HDD/SSD, and NVMe SSD are supported.
• On each storage device, at least two block channels (per controller) must be dedicated to the internal network. The channels can be Fibre Channels (FC) or Ethernet channels.
• On each storage device, two block channels (per controller) must be selected for the connection with the witness server. The channels must be Ethernet channels.

  Note:

  It is highly recommended that you choose different channels for the internal network and the witness network, unless there are not enough channels available.

• Additional host boards will be needed for the HA service.
• The witness server must meet the following hardware requirements.
  • Memory: 4GB DDR or more
  • Storage capacity: 8GB or more
  • Connectivity: Ethernet. Witness server link supports IPv4 only.
  • The witness server software can be installed as application software on a virtual machine or a physical server with any of the following supported host OS: Windows / Linux / VMware / macOS - OS X El Capitan 10.11 or later
• It is recommended that the witness server be deployed at a separately site from the storage devices.

Firmware & Software Requirements

• Both storage devices must be installed with firmware that supports the HA service. Block-level HA service and file-level HA service are available on the respective firmware versions as shown below:

  Models	Block-level HA service	File-level HA service
  GS/GSe family (G3 series)	1.64A.21 and later	1.64N.25 and later
  GS/GSe family (other series)		1.64K.25 and later

  Important!

  For more details about updating the firmware, refer to the article below.
  How do I update the firmware on my system?

• To enable block-level HA service / file-level HA service, additional purchase of the corresponding license is required.
• Both storage devices must have an HA service license.
• The witness server must be installed with Infortrend’s Witness Server Setup Program, available at Downloads and on the software CD.

Network Requirements

• The RTT of the internal network must be less than 250 ms.

Limitations

• HA service and scale-out cannot be activated on the same system at the same time.
• To use block-level HA service and file-level HA service at the same time, you must use the same two storage devices and pair them with each other.
• HA volumes cannot be created on a symmetric Active/Active pool or a pool with storage tiering.
• Block-level HA volumes do not support the following features: Block-level cloud gateway, volume snapshot, replication pair (volume mirror/volume copy), and scheduled tasks (scheduled snapshot/replication/volume copy).
• For file-level HA service, LDAP server and EonCloud will not be taken over after a failover.
• For specific SAS HDD models with pre-configured U.2 NVMe SSDs, when file-level HA service is successfully enabled, the system only supports read cache.