A VxRail stretched cluster with VMware vSAN spans two physical sites.
As a best practice, evenly distribute the nodes in the cluster between the two physical sites. Use the third site to host the VMware vSAN witness appliance. The two data sites are connected using a high bandwidth or low latency network link. The witness site is connected to both data sites using lower bandwidth or higher latency network links.
The following features are available with the VMware vSAN witness appliance:
Deployed at a third site and is connected to both data sites.
Monitors the health of the VMware vSAN data store and ensures that the cluster remains available even if a site or network failure occurs.
The VxRail stretched cluster with VMware vSAN can continue to operate if hosts are available, even if there is a failure at one of the data sites. The witness host plays a crucial role in this process, as it ensures that the cluster quorum is maintained and that data is accessible even if a site fails.
Figure 1. VxRail stretched cluster with VMware vSAN
The stretched cluster provides the following features:
Increased resilience and availability for critical workloads. Enables organizations to distribute their data across multiple sites and protect against site-level failures.
Creates a stretched cluster between two geographically separate sites and synchronously replicating I/O between sites.
Tolerates the failure of an entire site.
Extends the concept of fault domains to data center awareness domains.
The following terms are used for VMware vSAN stretched clusters:
Preferred or primary site
Secondary site
Witness host
A dedicated VMware ESXi host or VMware vSAN witness appliance can be used as the witness host.
If the network is lost between the data sites, the witness components are stored on the witness host and provide a quorum to prevent a split-brain scenario.
This witness host is the third fault domain.
Table 1. VMware vSAN storage policies impacting stretched clustersThe following table provides VMware vSAN storage policies that impact stretched clusters:
Storage policy
Description
Dual site mirroring (stretched cluster)
Enables protection across sites.
None
Data on preferred (stretched cluster)
Keep data on primary site only, no cross-site protection.
None
Data on nonpreferred (stretched cluster)
Keep data on the secondary site only, no cross-site protection.
Primary Failures to Tolerate (PFTT/FTT)
Defines how many disk or node failures can be tolerated for each site. For stretched cluster, it is 2n+1 (n is the number to tolerate). For erasure coding, it is 4 or 6 (1 or 2 failures respectively). All-Flash is required for erasure coding (RAID 5 or 6).
Secondary Failures to Tolerate (SFTT) (stretched cluster)
Defines the number of nodes, disks objects where failure can be tolerated within a site.
Failure Tolerance Method (FTM)
Defines the RAID 1 Mirroring and RAID 5 or 6 and Erasure Coding selection of RAID selection.
Data is not available for the Topic
Please provide ratings (1-5 stars).
Please provide ratings (1-5 stars).
Please provide ratings (1-5 stars).
Please select whether the article was helpful or not.
Comments cannot contain these special characters: <>()\