There was a discussion yesterday on our Socialcast system. The question was what the max amount of FT virtual machines was and what dictated this. Of course there are many things that will be a constraint when it comes to FT (memory reservations, bandwidth etc) but the one thing that stands out and not many realize is that the amount of FT virtual machines per host is limited to 4 by default.
This is currently controlled by a vSphere HA advanced setting called “das.maxftvmsperhost”. By default this setting is configured to 4. This advanced setting is an HA advanced setting (in combination with vSphere DRS) and defines the max amount of FT virtual machines, either primary or secondary or a combination of both, that can run on a single host. So if for whatever reason you want a max of 6 you will need to add this advanced setting with a value of 6.
I do not recommend changing this however, FT is a fairly heavy process and in most environments 4 is the recommended value.
ok, in case If I have cluster with 2 hosts, that means I can run 2 FT VMs on each server?
host 1: FT-VM1, FT-VM2, Secondary-FT-VM3, Secondary-FT-VM4
host 2: FT-VM3, FT-VM4, Secondary-FT-VM1, Secondary-FT-VM4?
that means 4?
Michael Ryom says
But isn’t there a better way to calculate the real max… I have a hard time believing that a quad core single cpu with 2 1gig nics and 8gig memory server, can handle the same amount as a 12 core quad cpu 2TB memory, quad 10gig nics… ???