FT Limitation

[baber]

i read below limits about FT:

1 - The maximum number of fault tolerance vms allowed on a host in the cluster Both primary vms and secondary vms count towards this limit .the default value is 4

Is that means maximum vm that can FT protected on an esxi host is 4vms with 1vcpu?

2- Maximum number of vcpu aggregated across all fault tolerance vms on a host vcpu from both primary vms and secondary vms count toward this limit . The default value is 8

Is that means when we enable FT protected on vm1 that ave 4vcpu that's shadow (secondary vm) will be create on the other host with 4vcpu

Are my imagine from FT correct ?

BR

[rcporto]

1 - The maximum number of fault tolerance vms allowed on a host in the cluster Both primary vms and secondary vms count towards this limit .the default value is 4

Is that means maximum vm that can FT protected on an esxi host is 4vms with 1vcpu?

No, that means that the maximum number of FT VMs (primary or secondary) is 4 and sum of vCPU should be no more than 8 vCPU. For instance, you can have 4 Primary VMs with 2 vCPUs each, or another scenario is to have 2 Primary VMs and 2 Secondary VMs, each VM with up to 2 vCPUs.

2- Maximum number of vcpu aggregated across all fault tolerance vms on a host vcpu from both primary vms and secondary vms count toward this limit . The default value is 8

Is that means when we enable FT protected on vm1 that ave 4vcpu that's shadow (secondary vm) will be create on the other host with 4vcpu

The shadow VMs will always have the exact configuration of the primary VM. But what the item #2 says is that the sum of FT VMs running on host should be at maximum of 8. For instance, you can have 2 Primary VMs and 2 Secondary VMs, each VM with up to 2 vCPUs running on a particular host, and the total number of vCPU in use by FT VMs are 8.

[baber]

thanks but confused

1- beforehand is that correct max number of vcpu that a vm can support FT is 4vcpu ?

2- is your means from primary = protected vm and secondary = shadow of protected vm ?

3- you said ' the maximum number of FT VMs (primary or secondary) is 4 and sum of vCPU should be no more than 8 vCPU '

is that your means just 1vm with 4vcpu or 2vm with 2vcpu or 4vm with 1vcpu can be put on an esxi and Protected?

BR

[rcporto · Accepted solution]

1- beforehand is that correct max number of vcpu that a vm can support FT is 4vcpu ?

Yes, see:

2- is your means from primary = protected vm and secondary = shadow of protected vm ?

Yes.

3- you said ' the maximum number of FT VMs (primary or secondary) is 4 and sum of vCPU should be no more than 8 vCPU '

is that your means just 1vm with 4vcpu or 2vm with 2vcpu or 4vm with 1vcpu can be put on an esxi and Protected?

No, you can can 2 VMs with 4 vCPUs, or 4 VMs with 2 vCPU... and VMs can be either primary (protected) or secondary (shadow)... this is just math... the maximum number of VMs is 4 and the sum of vCPU is 8, no matter how you configure your VMs, of course you can't assign more than 4 vCPU per VM.

[baber]

so thanks

i understand we can start maximum 4 vms on a host with total 8vcpu

now want to know if i had a cluster with 8 esxi host is that your means totaly can run 32vms with 64vcpu ?

BR

Babak

[baber]

can anyone help me ?

[rcporto]

now want to know if i had a cluster with 8 esxi host is that your means totaly can run 32vms with 64vcpu ?

Yes, you are correct, just keep in mind that the VMs can be either primary (protected) or secondary (shadow).

[baber]

so thanks

but imagine i have a cluster with 4 esxi hosts now in this mode when we say can run 4 vms on each host do i have to regard secondary vms or not ?

the other means totaly can run 16vms for FT in this cluster  or 8 primary vms + 8 secondary vms ?

BR

[baber]

Can you help me ?

[rcporto]

You should always consider the secondary VMs, then you can have maximum of 8 primary VMs (and 8 secondary) assuming your VMs have 2 x vCPU.

Here is an example of VMs distribution:

Green squares are Primary VMs and yellow squares are the Secondary VMs.

[baber]

is your means actually on each host can enable FT on 2 vms that each of them had maximun 2vcpu ?

BR

Babak

[a_p_]

The point is that the independent limits are either max. 4 VMs, or max. 8 vCPUs, where both Primary VMs and Secondary VMs count toward these limits.

So in case of 4 FT VMs with 2 vCPUs each on a single host, possible combinations are: 1 primary + 3 secondary VMs, 2 primary + 2 secondary VMs, or 3 primary + 1 secondary VM.

André

[baber]

excuseme i am so confused you said FT on a host will be support 4vms with 8vcpu now i understand your means from 4vms i primary vms+secondary vms

but my question here :  secondary  vm can not up on the same host that primary vm has been therby can enable FT on 2vms that each of them has 4vcpu on host1 ?

[a_p_]

secondary  vm can not up on the same host that primary vm has been ...

That's correct, a secondary VM will run on another host. Otherwise FT wouldn't make much sense, would it?

... therby can enable FT on 2vms that each of them has 4vcpu on host1 ?

Also correct. However, keep in mint that the secondary VMs for these 2 VMs will count against the FT limits on the other host(s) on which they will reside.

André

[baber]

is that means when i enable FT on 2 vms on host1 that each of them has 4vcpu  secondary vm will be create on host2 now therby we could not Enable FT on host2 for any vm

is that correct ?

BR

[a_p_]

Yes, if both secondary VMs are created on a single host, then the FT limit on this host is reached.

André

[baber]

so thanks

i understand these :

1- for example if i have 5 esxi host finally i can just enable FT for 2 vms on each esxi host with maximum 2 vcpu for each vm if secondary vms for host 1 create on host 2 and secondary vms for host2 created on host1 and ...

therby i have maximum 10 vms that protected by  FT in my cluster that  maximum vcpu can be used is 2vcpu for each of them and can not use 4vcpu for each of them because when we say each host can support maximum 4vms with 8vcpu (primary+secondaty) now on host1 create 2vms with each of them 2vcpu now secondary vms for these will be create on host2 (now on host 2 we have 2 vms with 4 vcpu totally) now if enable FT for 2 vms on host2 with 2vcpu secondary vms will be create on host1 now on host 1 we have 2 primary vms with 4vcpu and with 2 secondaru vms with 4 vcpu

is that correct ?

2- and in vmware documents i read said  we need minimum 3 esxi host for FT but we can run FT with 2 host is that means it said 3 hosts because if one of them has been failed our primary or secondary moved on the other ?

BR

Babak

[baber]

is that correct ?

BR

[a_p_]

I hope you had some great Easter Holidays too.

1. That's basically correct. You can power on FT protected VMs until host maximums are reached. However, keep in mind that there's an automatic placement for secondary VMs, and that you may want to plan with sufficient resources to allow FT do its job, i.e. create a new secondary VM in case a primary VM fails, and a failover occurs.
   for more details see FAQs at  https://kb.vmware.com/s/article/1013428​
2. With just two hosts, FT will not be able to create another secondary VM in case of a failure. This can basically be compared with a RAID1 with, and without a Hot-Spare disk.

André

[baber]

I hope you had some great Easter Holidays too.

That's basically correct. You can power on FT protected VMs until host maximums are reached. However, keep in mind that there's an automatic placement for secondary VMs, and that you may want to plan with sufficient resources to allow FT do its job, i.e. create a new secondary VM in case a primary VM fails, and a failover occurs.

for more details see FAQs at  https://kb.vmware.com/s/article/1013428

imagine i have 3 host in my cluster if i protect 2vms on host1 and 2 vms on host and 2 vms on host3 in this mode maximums are reached and can not enable FT for the other vms

is that correct ?

With just two hosts, FT will not be able to create another secondary VM in case of a failure. This can basically be compared with a RAID1 with, and without a Hot-Spare disk.

is this vmware best practice that if want enable FT on a vm use minimum 3 hosts ?