Availability and redundancy
Google Cloud VMware Engine provides availability for your applications running on VMware in your private cloud environment. The following table lists failure scenarios and the associated availability features that help protect your applications.
|Failure scenario||Application protected?||VMware Engine HA feature||VMware HA feature||Google Cloud feature|
|Disk failure||Yes||Fast replacement of failed node||About the vSAN Default Storage Policy|
|Fan failure||Yes||Redundant fans, fast replacement of failed node|
|NIC failure||Yes||Redundant NIC, fast replacement of failed node|
|Host power failure||Yes||Redundant power supply|
|ESXi host failure||Yes||Fast replacement of failed node||VMware vSphere High Availability|
|VM failure||Yes||VMware vSphere High Availability||Load balancer for stateless VMware VMs|
|Leaf switch port failure||Yes||Redundant NIC|
|Leaf switch failure||Yes||Redundant leaf switches|
|Rack failure||Yes||Placement groups|
|Network connectivity to on-premises||Yes||Redundant networking services||Redundant Dedicated Interconnect and Partner Interconnect circuits|
|Network connectivity||Yes||Redundant Dedicated Interconnect and Partner Interconnect circuits|
|Regional failure||Yes||Hosting regions|
Fast replacement of a failed node
VMware Engine continuously monitors the health of VMware clusters. When VMware Engine detects an ESXi node failure, it adds a new ESXi host to the affected VMware cluster from its pool of readily available nodes and removes the failed node from the cluster. This functionality quickly restores the spare capacity in the VMware cluster, supporting the cluster's resiliency provided by vSAN and VMware vSphere High Availability (HA).
A user who creates a private cloud can select a region and a placement group within the selected region. A placement group is a set of nodes spread across multiple racks but within the same spine network segment. Nodes within the same placement group can reach each other with a maximum of two extra switch hops. A placement group is always within a single availability zone and spans multiple racks. The control plane distributes nodes of a private cloud across multiple racks based on best effort. Nodes in different placement groups are guaranteed to be placed in different racks.
VMware Engine private clouds are hosted in a user-selected Google Cloud location. These locations are composed of regions and zones. A region is a specific geographical location where you can host your resources. Each region has one or more zones.
Resources in different regions are isolated from most types of physical infrastructure and infrastructure software service failures. You can design a robust system and distribute resources across different failure domains to protect your applications and data from data center failures.
Redundant networking services
All Google networking services for the private cloud (including firewall, public IP addresses, internet, Dedicated Interconnect, Partner Interconnect, and Cloud VPN) are highly available and support the SLA.