Abstract: A method for a secondary host to support continuous availability for an application on a primary virtual machine on a primary host is disclosed. The method includes the secondary host creating a secondary virtual machine that is identical to the primary virtual machine, the secondary host receiving activities of the primary virtual machine from the primary host, the secondary host buffering the activities, and the secondary host determining if the buffered activities are safe to replay. When the buffered activities are determined to be safe to replay, the method includes the secondary host replaying the buffered activities to the secondary virtual machine. When the buffered activities are determined to be unsafe to replay, the method includes the secondary host discarding the buffered activities and setting the secondary virtual machine as a new primary virtual machine to take over a service provided by the application.

Abstract: A balancer can balance workloads in an active-active or active-passive stretched cluster by determining where to provision active instances of virtual machines. The balancer can maintain a set of compatibility rules to govern which hosts, such as sites or servers, are compatible with which instances. The compatibility rules can ensure that the same host does not run both instances, for failover purposes. Additionally, the balancer can create a key-value map that tracks performance of various available hosts. By monitoring host performance and maintaining the compatibility rules, the balancer can re-provision instances on different compatible hosts when performance of a current host suffers.

Abstract: Techniques for automatic rule based grouping of compute nodes for a global optimal cluster are disclosed. In one embodiment, a virtual infrastructure management (VIM) server may obtain a list of operating, provisioned and/or about to be provisioned compute nodes in a cluster. The VIM server may then obtain user configured and run-time network parameters of a network interface card (NIC) of each compute node in the list. Further, the VIM server may then measure round-trip times (RTTs), using a ping, between each compute node and each of remaining compute nodes in the list. Furthermore, the VIM server may then group the compute nodes in the list based on the obtained user configured and run-time network parameters and/or the measured RTTs. In addition, the VIM server may perform a high availability (HA) operation in the cluster using the grouped compute nodes.

Abstract: A balancer can balance workloads in an active-active or active-passive stretched cluster by determining where to provision active instances of virtual machines. The balancer can maintain a set of compatibility rules to govern which hosts, such as sites or servers, are compatible with which instances. The compatibility rules can ensure that the same host does not run both instances, for failover purposes. Additionally, the balancer can create a key-value map that tracks performance of various available hosts. By monitoring host performance and maintaining the compatibility rules, the balancer can re-provision instances on different compatible hosts when performance of a current host suffers.

Abstract: Techniques for automatic rule based grouping of compute nodes for a global optimal cluster are disclosed. In one embodiment, a virtual infrastructure management (VIM) server may obtain a list of operating, provisioned and/or about to be provisioned compute nodes in a cluster. The VIM server may then obtain user configured and run-time network parameters of a network interface card (NIC) of each compute node in the list. Further, the VIM server may then measure round-trip times (RTTs), using a ping, between each compute node and each of remaining compute nodes in the list. Furthermore, the VIM server may then group the compute nodes in the list based on the obtained user configured and run-time network parameters and/or the measured RTTs. In addition, the VIM server may perform a high availability (HA) operation in the cluster using the grouped compute nodes.

Abstract: Failover processing of a service in a cluster of nodes is disclosed. A node brings up a failover instance of a service in response to an indication of a failure in an initial instance of the service. The failover service executes in the background so as not to be visible to users and thus does not receive user requests. The background service is terminated in response to the failure indication being a false positive. The background service is executed in the foreground in response to detecting that the initial instance of the service has failed and processes user requests.

Abstract: System and method for providing failover protection for a cluster of host computers operates in one of two modes. When the number of host computers in the cluster is equal to or less than a threshold number, the host computers in the cluster operate in an all-active heartbeating mode in which each of the host computers is enabled to function as a master node. When the number of host computers in the cluster is greater than the threshold number, the host computers in the cluster operate in a distributed heartbeating mode in which some of the host computers function as master nodes and at least one of the host computers functions as a slave node.

Abstract: Failover processing of a service in a cluster of nodes is disclosed. A node brings up a failover instance of a service in response to an indication of a failure in an initial instance of the service. The failover service executes in the background so as not to be visible to users and thus does not receive user requests. The background service is terminated in response to the failure indication being a false positive. The background service is executed in the foreground in response to detecting that the initial instance of the service has failed and processes user requests.

Abstract: A method for a secondary host to support continuous availability for an application on a primary virtual machine on a primary host is disclosed. The method includes the secondary host creating a secondary virtual machine that is identical to the primary virtual machine, the secondary host receiving activities of the primary virtual machine from the primary host, the secondary host buffering the activities, and the secondary host determining if the buffered activities are safe to replay. When the buffered activities are determined to be safe to replay, the method includes the secondary host replaying the buffered activities to the secondary virtual machine. When the buffered activities are determined to be unsafe to replay, the method includes the secondary host discarding the buffered activities and setting the secondary virtual machine as a new primary virtual machine to take over a service provided by the application.

Abstract: System and method for providing failover protection for a cluster of host computers operates in one of two modes. When the number of host computers in the cluster is equal to or less than a threshold number, the host computers in the cluster operate in an all-active heartbeating mode in which each of the host computers is enabled to function as a master node. When the number of host computers in the cluster is greater than the threshold number, the host computers in the cluster operate in a distributed heartbeating mode in which some of the host computers function as master nodes and at least one of the host computers functions as a slave node.