VC Host Memory Usage vs esxtop SZTGT

[UoGsys]

Hi All,

This has come up on the back of a recent case logged to VMware support.

Has anyone else noticed an inconsistency between the above stats reported by VC for each host, against the one's reported from the esxtop command? For example, at a point in time The SZTGT stats is what the vmkernel gives each VM, which comes to around 19GB. But if you check the host memory usage in VIC, this says about 13.5GB - which doesn’t add up and make sense‼!

The worrying thing is doesn't VC base DRS decisions on the host memory usage (as well as others stats)? Also, in the case of memory over-commit, wont VC base this on its stats, resulting in ballooning kicking in for the guest VM's running on a host?

Any views or am I talking total rubbish!!!??!!!

[depping]

SZTGT also includes all overhead, could this be the difference you're seeing?

Duncan

VMware Communities User Moderator | VCP | VCDX

-

---

Blogging: http://www.yellow-bricks.com

Twitter:

If you find this information useful, please award points for "correct" or "helpful".

[UoGsys]

Thanks depping - if this includes all overhead on the host then shouldn't this be the stat that is reported by VC? My concern here is that we base spare host capacity on these stats over a period of time. If the VC stat does not report the full host memory usage then ESX could be swapping without us being aware of it?