agsandrew - Fotolia
Fault Tolerance is an emblematic feature within VMware vSphere that provides continuous availability for a virtual machine by creating a duplicate of that VM. Since both copies are kept in sync, if the primary copy fails, the secondary VM will take its place, preventing downtime.
VMware Fault Tolerance (FT) protects VMs without any additional software on the VM or host. When FT was first released, many users were impressed by the simple ability to right-click and enable it on a per-VM basis. However, latency between the primary and secondary VMs created a problem. In some cases, users were unable to use VMware FT on latency sensitive applications. The release of vSphere 6 reduced latency significantly, and the latest version of vSphere, 6.5, reduces latency even more.
Fault Tolerance has a new engine
Instead of the traditional vLockstep technology, the updated version of FT packaged in vSphere 6.5 uses Fast Checkpointing, which helps keep the primary and secondary VMs in sync with snapshots.
The latest version of VMware FT still requires 10 Gigabit Ethernet and you can now use multiple port groups to increase the overall bandwidth for vSphere FT traffic logging. Providing multiple lanes for FT traffic reduces latency because it allows FT-enabled VMS to do a lot of work in-memory.
VMware FT gets smart
When you activate FT in vSphere 6.5, it will prompt you to choose a location for the secondary VM. You can determine the best location for the secondary VM by ranking ESXi hosts based on available network bandwidth. There are two types of ranking on which to base your decision: network ranking, which checks network throughput, and data store ranking, which ranks data stores according to certain criteria such as space available on each data store. You can also override the hard requirements concerning the maximum number of FT VMs with the advanced VMware High Availability configuration. Use the following commands to set maximums for VMs and vCPUs.
das.maxftvmsperhost -- This value is maximum number of fault tolerant VMs allowed on a host in the cluster. Both Primary VMs and Secondary VMs count toward this limit. The default value is 4.
das.maxftvcpusperhost -- This value is maximum number of vCPUs aggregated across all fault tolerant VMs on a host. VCPUs from both Primary VMs and Secondary VMs count toward this limit. The default value is 8.
Changes to licensing
In addition to the aforementioned changes, vSphere 6.5 introduces a licensing limit on the number of supported FT VMs. The vSphere Standard and Enterprise licenses allow up to two vCPUs, while the vSphere Enterprise Plus license allows you to have up to four vCPUs.
Note that VMware FT is not supported in vSphere Essentials and vSphere Essentials Plus kits.
Improve VMware Fault Tolerance in the data center
What are the VMware Fault Tolerance requirements?
VMware FT gets a makeover in vSphere