Книга: Mastering VMware® Infrastructure3

vReplicator

vReplicator

vReplicator, previously known as esxReplicator, is a host-level image replication solution for VMware infrastructure. Also developed by Vizioncore Inc., vReplicator is a Windows-based application that enables companies to leverage virtualization technology to keep applications running day in and day out all year round by replicating entire virtual machines, including configuration settings, patches to the OS, the application and the data, and all other OS-level changes.

vReplicator selectively replicates virtual machines from many hosts to one host, as well as to dissimilar hardware platforms. In physical server environments, disaster-recovery solutions are expensive and difficult to deploy and maintain because they require redundant, overbuilt standby servers that are typically dormant and underutilized until a portion of the production environment becomes unavailable. Virtualization eliminates the expense of overbuilding by allowing higher utilization of existing servers. vReplicator offers a strategic and bandwidth-conscious approach to replication: it enables full virtual machines to be selectively and cost-effectively replicated without agents, and with full flexibility to complement the existing server environment, while reducing the financial impact of the overall solution.

vReplicator selectively replicates individual virtual machines selected by the virtual infrastructure administrator. This eliminates the maintenance, network traffic, and hardware and software expenses of replicating all the LUNs from a SAN. Because vReplicator works from a centralized server, installation components are not needed on either the source or the target virtual machines.

vReplicator operates from a centralized server and does not need to be installed on either the source or the target virtual machines, which eliminates the licensing costs, management, and maintenance involved with agent-based solutions. vReplicator provides a Setup wizard for the configuration of replication jobs. The solution can be configured for direct communication between the vReplicator server, the source, and the destination hosts, or it can be configured to leverage VirtualCenter. Replication job settings include replication intervals, the destination server, and the target datastore. Status information is available in the GUI to determine the duration of replication processes as well as to verify successful replications. Performance statistics are available to determine the workload on a host and the impact of replication on the virtualized environment. These statistics are helpful in determining the limitations of the host and the performance requirements for replication. vReplicator is integrated with VirtualCenter to provide a recognizable hierarchy of virtual machines and the respective hosts involved in replication. This integration with VirtualCenter allows the solution to be VMotion and DRS aware so replication jobs continue even after the virtual machines have moved.

The initial synchronization for vReplicator leverages proven Vizioncore technology for the initial copy of the source virtual machine. After the initial synchronization, a differential engine sends only changes to the destination host and applies them to the closed target virtual machine. The replication engine of vReplicator is a Windows service that resides on the server where vReplicator has been installed. This reduces the impact on the underlying virtualization platform implementation by offloading the work to the dedicated vReplicator server. vReplicator validates the differential information needed to perform live replication without interruption to the virtual machine. The differential engine checks for changes to VMDKs as a catalyst for performing block-level replication. To prevent problems in the replication and recovery process, vReplicator copies changed blocks to the target system before applying them. By copying the blocks before writing them into the replicated virtual machine files, vReplicator ensures that if any problems arise during the process the original blocks can be written back to the target. Ultimately this logic prevents the need to recopy the entire VMDK.

Оглавление книги