Workstation – Page 7

Quick Tip - How to enable memory overcommitment in VMware Fusion?

02.14.2015 by William Lam // 20 Comments

There was an interesting internal thread that I came across yesterday where someone was asking if VMware Fusion supported memory overcommitment like VMware Workstation? In VMware Workstation, memory overcommitment can be adjusted by going to Edit->Preferences->Memory which provides three different options as shown in the screenshot below. This setting specifies the percentage of memory that would be reserved from the physical host memory for each Virtual Machine.

These options map to the following % of host memory reservation:

Option	% of Host Memory Reserved Per VM
Fit all virtual machine memory into reserved host RAM	100
Allow some virtual memory to be swapped	50
Allow most virtual memory to be swapped	25

Disclaimer: Be aware, that default safe guards have been put in place to ensure optimal VM performance. If you decide to change these settings and allow memory overcommitment, it can potentially degrade performance of both your VMs as well as host system. Make sure you understand the changes before applying them.

You can also specify a custom value by editing the VMware Workstation configuration file located in: C:\ProgramData\VMware\VMware Workstation\config.ini and modifying or adding the following property:

prefvmx.minVmMemPct = P

where P is the percentage of configured VM memory that should fit into the host memory. The smallest value that P can be is 1. Below is a screenshot of a 32GB VM running on a Macbook Air with latest version of Fusion (8.5.3) which only has 8GB of physical memory and the value that I had used for this demonstration is 1.

Going back to VMware Fusion, memory overcommitment is also possible but the option to configure it is not available in the VMware Fusion UI. You will need to add the above setting into the VMware Fusion configuration file located in /Library/Preferences/VMware\ Fusion/config which does not exist by default. You will need to restart Fusion/Workstation for the change to go into effect.

In addition to the change, if you do decide to overcommit your memory, it was also mentioned that you may also want to disable Hard Disk buffering for optimal performance. You can make this change in the Advanced Settings of the the VM as seen in the screenshot below.

You can also just add the following property to the Virtual Machine's VMX configuration file:

hard-disk.hostBuffer = "disabled"

Memory commitment can be a wonderful tool, especially for lab environments. If you combine this with SSD storage and if swapping does occur, the impact may be acceptable so that you can run a few more VMs. Thanks to Regis Duchesne & Jesse Pool for sharing this handy tidbit!

Ultimate automation guide to deploying VCSA 6.0 Part 0

02.09.2015 by William Lam // 8 Comments

With vSphere 6.0, there is a new deployment model for vCenter Server which is comprised of following two core components:

Platform Services Controller (PSC) Node - Provides VMware Infrastructure services such as vCenter Single Sign-On, vSphere Licensing and VMware Certificate Authority Management (VCMA)
vCenter Server Management Node - Provides vCenter Server Service, Inventory Service, vSphere Web Client, vPostgres DB, vSphere Syslog Collector, vSphere Auto Deploy, and vSphere Dump Collector Services

From these two components, there are three deployment types (also shown in the diagrams below):

Embedded Node - Both the Platform Services Controller and vCenter Server Management Node reside on a single system, this is true for both the Windows vCenter Server and the VCSA
External Platform Services Controller Node - You can deploy multiple PSC's and configure them with independent SSO Domains or have them all joined to a single SSO Domain, replicating between each other
vCenter Server Management Node - This requires that you have deployed an external PSC which the vCenter Server can point to

There are currently two supported methods of deploying the VCSA 6.0 Appliance which is using the new HTML based UI (Supported only on Windows) or a new scripted installer method (supports Windows, Mac & Linux). Both of these methods today require direct access to an ESXi host for deployment, which may not work for everyone. What if you want to deploy the new VCSA 6.0 using an existing vCenter Server or running it on top of VMware Fusion or Workstation? Luckily, I spent quite a bit of time going through all these "alternative" deployment methods and documenting the process so that you have a choice on how you would like to test and evaluate vSphere 6 and the new VCSA in your environment.

These alternative methods will be using the VCSA OVA which is actually included in the VCSA ISO. You will need to extract the contents of the VCSA ISO and you can find the OVA in the following path after extraction: VMware-VCSA-all-6.0.0-2562643->vcsa->vmware-vcsa where vmware-vcsa is the VCSA OVA file. Depending on the deployment method you are using, you may only need to just extract the contents of the ISO or possibly rename the vmware-vcsa with .ova extension to deploy. Please refer to the articles below for more details.

Disclaimer: Though these alternative deployment options work, they are however not officially supported by VMware. Please use at your own risk.

In the upcoming days, I will be sharing a 4-part blog series for automating the deployment of the new VCSA 6.0 with the following deployment options:

In each article, I will provide resources on how to deploy to an existing vCenter Server or directly to an ESXi host using ovftool via a shell script as well using PowerCLI, deploying to VMware Fusion and deploying to VMware Workstation. Stay tune for Part 1 ...

CoreOS is now available as OVA in Alpha channel

01.08.2015 by William Lam // 11 Comments

It looks like the folks over at CoreOS have now also produced an OVA image which can be easily imported into a vSphere or even vCloud Air environment. Previously, it took a few addition steps to convert the "hosted" disk image originally meant for VMware Fusion/Workstation to properly work in a vSphere/vCloud Air based environment. The CoreOS OVA is currently only available in the CoreOS Alpha channel for the "Production" image which also includes VMware's open-vm-tools and the latest release as of today is CoreOS 554.0.0.

You can using either the vSphere C# or vSphere Web Client to import the OVA or you can automate this simply by using command-line via ovftool. Here's an example snippet that you can run directly against an ESXi host:

/Applications/VMware\ OVF\ Tool/ovftool \
        --name=CoreOS \
        "--net:VM Network=VM Network" \
        --datastore=mini-local-datastore-2 \
        --diskMode=thin \
        'http://alpha.release.core-os.net/amd64-usr/554.0.0/coreos_production_vmware_ova.ova' \
        'vi://root:*protected email*'

You can also import the CoreOS OVA into vCloud Air's but you will need to connect into the vCloud Director interface to upload or you can also use ovftool. For more details on how to import using ovftool, check out their documentation here.

Here's a screenshot of deploying CoreOS from a vCloud Air Catalog:

The "Production" CoreOS image does not contain insecure SSH keys as the "insecure" image and so you will still need to create a Cloud Config ISO if you wish to further customize the image including login credentials. You can take a look at the script I had created for deploying CoreOS from the Stable channel and for more details check out the Cloud Config documentation as well.