31 Jul 2013

VMware vLockstep

VMware vLockstep is technology that captures inputs and events that occur on a primary virtual machine (VM) and sends them to a secondary VM. VMware vLockstep is the technology that supports VMware's Fault Tolerance component of VMware vSphere.

VMware's Fault Tolerance works by keeping a primary virtual machine (VM) and a secondary VM in perfect sync. VMware vLockstep captures inputs and events that occur on the primary VM and sends them to the secondary VM. Because the secondary VM is always in sync with the primary VM, it can take over in the event of a primary VM failure without interruption and provide fault tolerant protection. When the secondary VM takes over, VMware FT automatically creates a new secondary VM. The name lockstep comes from a style of military march that emphasizes synchronous movement.

VMware vLockstep should be set up on a dedicated network interface card (NIC) with at least 1 GB. Although all data is synchronized between the paired VMs over a server backbone network, outputs are suppressed in the secondary VM. For instance, VMware FT ensures only the primary VM initiates write operations to storage. Certain actions and instructions that are irrelevant for the secondary VM are not synched via vLockstep, reducing the burden on disk space and processors.

For vLockstep to reproduce CPU instructions from the primary VM on the secondary VM, the Intel or AMD processors used must have the appropriate performance counter architecture and virtualization hardware assists. Both hosts supporting the VM pair must be in the same processor family.

In versions of vSphere earlier than v.5, the vLockstep VM pairs were marked "disabled" in VMware Distributed Resource Scheduler (DRS), enabling higher compatibility between VMware FT and DRS.

No comments:

Post a Comment