Solved

VMware ESXi Best Practices

  • 10 November 2015
  • 3 replies
  • 4007 views

Userlevel 1
Badge +6
Greetings community members!

I am brand new to the Nutanix platform and have completed my basic initial configuration of my nodes on ESXi 5.5U2, however I am having trouble finding (not through lack of Google-fu) a complete guide on what Nutanix recommends as BPs for configuring hosts for ESXi.

This raises some questions... e.g:

Should I add anything to my node's Advanced Settings?

Should I chanage the power settings Active Policy to High Performance?

What is the definitive guide to HA settings with respect to the nodes, VM monitoring, policies therein, etcetera ?

Not that this is an exhaustive list of questions, but I forgot to ask our Nutanix Pro Services rep before he left...and I could not immediately find what I need unless I am looking in the wrong place! :$

Edit: Some of what I asked is in the vSphere Administration Guide for Acropolis which is very helpful, but was curious if there is any additional documentation for Best Practices with Nutanix in vSphere. Apologies for any confusion.

Edit 2: vSphere Administration Guide for Acropolis contains the general guidelines needed for most purposes. I will keep an eye out for upcoming document releases in the support portal.

Thank you for reading, all.
icon

Best answer by Jon 12 November 2015, 15:38

View original

This topic has been closed for comments

3 replies

Userlevel 1
Badge +5
Hi 

The guides you found include most of the "recommended practices" that I'm personally aware of. A lot of it is what the particular hypervisor (in your case ESXi) recommends but there are some things to take into consideration regarding the CVM's and a few other things to optimize that are unique to Nutanix. This is not an all inclusive prescriptive list but some of the things I do / take into consideration when setting up a Nutanix vSphere cluster are:

vSphere HA Settings
- Enable host monitoring
- Enable admission control
- Disable VM restart priority of all CVMs
- Set the host Isolation Response of all CVMs to "leave powered on"
- Disable VM Monitoring for all CVM's
- Enable Datastore Heartbeating and choose Nutanix NFS
- Add "das.ignoreInsufficientHbDatastore=true" in the advanced settings (relates to the previous datastore heartbeating setting since there is probably only one datastore)

vSphere DRS Settings
- Disable automation on all CVMs
- Leave power management disabled/off
- Automation enabled at cluster level with conservative setting (excessive vMotion impacts data locality, amount of data copied across the cluster if/when a VM is moved with DRS)
Userlevel 6
Badge +29

Who did you work with in PS? I'd be happy to get you guys re-connected (That's my team), and they can help out, as they'll know more about your environment than anyone on the open forums here.

That said, honestly, you dont have to configure much past the default, keep it simple principle for sure.

The only non-default things that MIGHT be in-scope would be these:
1 - esxi syslog
2 - network (distributing switching, etc)

RE Power Policy - leave it to balanced. ESXi does a really good job at handling C and P states, and honestly, it won't make much of a difference unless you have VERY cpu scheduling sensitive applications like HPC or vBCA.

Otherwise, defaults are pretty strong, and our foundation tool does a lot of the normal heavy lifting right out of the box, so it's definitely a lot less host config than a vanilla esx install by hand.

Userlevel 1
Badge +6

Sent you a PM.

Looks like you condensed the major points of the vSphere guide quite well in your reply. Much appreciated!