Breakfast
Nothing to see here
Lunch
Repeat from Monday (chicken, lasagna, polenta)
Sessions
TAM Vsphere web client roadmap
TAM VSAN hardware
TAM VSAN encryption
INF72827 – vSphere DRS Deep Dive: Understanding the Best Practices, Advanced Concepts, and Future Direction of DRS
This session was super packed with information. A quick overview of DRS led into a few best practices.
Use latency sensitive option for VMs to prevent t hem from vMotioning via DRS. They will only move during maintenance mode or HA. I have a pending question if this is better than disabling DRS/setting to manual for the individual mode. Most of the time I have problems with those latency sensitive VMs when I put a host into maintenance mode. In those cases I schedule an outage to finish the maintenance mode.
Set bios to os controlled and ESXi power management to high performance. I have been questioning this wisdom forever as I get different responses from different VMware employees and from other customers. I believe regardless of Bios overall setting, the deep c states (c1, c1e) should be enabled. Then maximum performance profile at bios or os controlled/high performance should be used.
There are a few Advanced options that can be set to force DRS to have different behavior. In vSphere.Next, those common options will be exposed in the gui:
VCPU to core – can set this to ensure a vcpu to core ratio is not exceeded
Balance based on active or consumed memory
Consumed is an area that has ever been touched
Active is the amount that is currently being used
There will also be DRS algorithim improvements that will smooth out workloads over the cluster as well as improve initial placement of vms PRIOR to power on. if there is a way to force reservations for all guests in cluster.
Nighttime
I grabbed a burger and called it a night. I was falling asleep in sessions and decided to take a nap.