Strange behave of DRS

[vMario156]

I am actualy experience a strange behavior of DRS, which I dont understand:

My cluster contains 10 ESXi 5 hosts, all the same version, same datastores, same networks etc.

I am using one DRS rule, which puts around 20 Win2k VMs togehter (VMs should run, not must) on just 2 of this hosts (to get most out of TPS).

vCenter is telling me that my cluster is immbalace (target is 0,023 but it is current at 0,071).

So far so good, now the strange thing:

On both hosts (which are used inside the rule to get all the 20 VMs with Win2k) are also some other VMs (2-4 per host), which could be moved by DRS to some other hosts. Even if I set DRS to the highest aggresitiv level (fully automated) its not moving any of this VMs. And there is no reason why this VMs are not moveable (manual it is possible, validation successfull).

Anyone any idea where to start troubleshooting or if theres anything I missing / got wrong ?

Cheers,

Mario

[weinstein5 · Accepted solution]

This is a common misconsption about DRS - DRS role is to insure delivery of resources to the VM - so if you have a 5 identical node cluster with 200 VMs you theoretically could have 100+ VMs on one host and the remaining spread across the remaining nodes as long as VMs were receiving the necessary resources.

The aggressiveness in the automation is an indication on what level of improvement the VM will see if it is migrates - in other Aggressive means VMs will be vmotioned even if there will ony be a slight improvement (lees than 5%) in delivery of resources.

[steffanhr]

The risk and cost of moving the VM is not justifiable compared to the possible gain of resources gained by moving the VM.

[vMario156]

thanks, of course u guys are right.

after i read your answers i remembered reading this a wile ago.

thanks!