Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

3270

March 8th, 2016 08:00

Support for Fault Tolerance on any VxRail model?

Hello,

I understand that the VxRail models will come with vSphere Standard Edition, which support 2 vCPU Fault Tolerance.  However Fault Tolerance requires its own 10 GigE connection, so does this mean that VMware Fault Tolerance is NOT supported on any VxRail model?

March 8th, 2016 08:00

Correct, not supported - VxRail is implementing other and better ways to achieve active-acvtive failover .  More detailed answers are in the Ask the Experts thread (starting on Page 5)

143 Posts

March 8th, 2016 09:00

Thanks.  I checked out that thread and I see that stretched clusters will be supported, which is nice, but that will rely on VMware HA rather than Fault Tolerance, correct?  VMware HA is great for almost every workload (about 2-3 minutes of downtime after server/site failure), but I was curious if there will be a "zero" downtime option similar to VMware Fault Tolerance with VxRail?  In other words if a customer has an application that requires "zero" downtime in the event of a node/server/appliance/site failure, will VxRail be able to provide this capability?  Or will the customer need to look at physical servers with additional 10 GigE interfaces in order to leverage "zero downtime" Fault Tolerance with VSAN?

January 25th, 2024 12:53

This is a very old thread but as the answer given back then is wrong, I want to set things straight.

Fault Tolerance (FT) is a technique where two ESXi Servers run a Virtual Machine, active active in memory and CPU using CPU lockstepping.

This functionality, true hardware failure resilience, is not new as it was introduced on Mainframes and Mini's long ago. But this is what we (and VMware) call "fault tolerance".

VMware FT has nothing what so ever to do with VMware HA or stretched clusters. HA and stretched clustering can re-start a VM in case of ESXi Host hardware failure. In the case of stretched clusters even in another datacenter. But the VM always "dies" when the host dies. It merely get's automatically restarted on another ESXi server in the cluster.
FT really does protect against hardware failure because the VM is running on two ESXi Servers at the same time as CPU instructions and memory perform the same tasks all the time. CPU lockstepping is what that is called.

The answer:

"Correct, not supported - VxRail is implementing other and better ways to achieve active-acvtive failover .  More detailed answers are in the Ask the Experts thread (starting on Page 5)"

is not correct. VXrail can do vSAN stretched clustering which only relates to storage. VMware HA is not related, but can be used together with, vSAN Stretched clustering. But this is not fault tolerance and the VM will need to be restarted as boyler05 writes.

As vSAN / VxRail is based on ESXi, the real FT feature can be used together with Stretched Clustering & HA. This provides actual resilience against hardware failures as the VM will keep running completely uninterrupted when a hosts crashes. It needs both FT + vSAN Stretched Clustering to offer true "Fault Tolerance".

Kind regards,

Steven Rodenburg

VMware vExpert / VxRail Tech-Lead

(edited)

No Events found!

Top