Example of a Medium sized ZebClient HA cluster deployed on Hypervisors.
In a Hypervisor deployment, it is important to consider the disk redundancy and node redundancy of the Cluster Size to be deployed. For a High Availability (HA) Hypervisor deployment this could be of less relevance but for a single Hypervisor or multiple without HA enabled this must be considered not to lose more than given redundancy stipulated in Table Redundancy for Cluster Size for disks and VM’s. For a Hypervisor deployment without HA we recommend deploying a ZebClient HA deployment, this is a HA cluster on two separate Hypervisor hosts making the ZebClient cluster HA, see Table HA Cluster and then use VM configuration backup for each Application VM, to make it possible to boot the failed VM’s on the second Hypervisor host.
Table Redundancy for Cluster Size
|Cluster Size||Node redundancy||Disk redundancy|
Table Cluster HA
|Cluster Size||Nodes No||Warm Disks||Hot Disks|
The number of Application VM are then scaled to support the performance specified by the use-case applications, from 1 to multiple. For a High Performance scale-out application use-case, a large number of Application nodes can be deployed to serve data access from multiple nodes, for parallel workloads. All ZebClient cluster data will then be accessible from each Application VM, with the first read operation having Warm tier performance and consecutive read operations supporting Hot tier performance.
Stateless means there is no record of previous data operation and each operation request has to be handled based entirely on information that comes with it. There is also no stored knowledge of or reference to past data operations.
All Application VM’s using ZebClient for data access is Stateless making scaling Application VM both up and down simple and safe when the workload is shifted up or down, or an Application is to be moved to a different host or location.
Both node, disk redundancy, and performance needs to be considered for the deployment on the Hypervisor. All default Cluster Size deployments of ZebClient in Table Cluster have both node and disk redundancy if deployed on separate nodes without a single point of failure. In a Hypervisor environment, VMs are not separate nodes, and if a Hypervisor host fails several VM’s will be affected and must be considered not to take the ZebClient cluster down. The recommendation is to always use a ZebClient HA deployment from Table Table Cluster HA when the Hypervisor is not part of a HA cluster.