Afternoon. a philosophical question came to mind a short while ago as a result of a conversation with a co-worker: when is the appropriate time to clone a vm? Deploying from template or using sysprep excluded, my opinion would be the creation of a test vm from the original which I agree that it is the easiest, best way to duplicate a vm for test if needed.
Another way I could ask this question: is if you build a production vm and prior to putting into production, do you clone the vm you are building for a test vm? Or do you build a test vm and clone it to a production vm?
Second philosophical question is do you clone when a vm is live/ in production? my answer is not if we can help it
Third philosophical question is: if you clone a vm for the purposes of upgrade testing, do you promote the clone to production afterward?
I was been bit in the past by a duplicate windows sid\MAC address\Licensing issue when doing this so this is what is causing me to revisit this.
either way, if you have any thoughts i would appreciate to hear them
When to clone a VM ?
The technic of working with clones can vary a lot in different scenarios - lets talk about some different scenarios first.
A: your company runs a webshop: during summer you sell pumps, during winter you sell moonboots. After every season the webshop production data will be transferred to your tax-department.
B: your company offers IT-classes. Every week you have a new group of students. So after the end of each week you destroy all productions VMs and start over wish new copies on mondays.
C: your company runs a helpdesk service. Every helpdesk support-user gets a fresh copy of the master when je joins the team. Every user gets ome disk for his personal longterm use. And as you have to support various products you supply additional disks with all the new details for the new product versions every few weeks or so.
In all 3 scenarios a skilled administrator creates a master - including the OS and all the additional tools.
When the functionality works as expected the admin adds a new user and prepares everything so that no license-problems, mac-address issues or other problems will stand in the way during the clone process.
For the production phase he adds 2 more disks: one for user-data and one for the production data.
After the end of the production phase you have 3 vmdks in each scenario: os-and-tools.vmdk, user-data.vmdk and production-data.vmdk.
In scenario A you keep the production-data.vmdk and send it to the tax-department. The other 2 vmdks can be thrown away.
In scenario B you discard all 3 vmdks - without having to inspect the 3 vmdks before destruction.
In scenario C you always keep the accumulated user-data.vmdk and provide new clones of os-and-tools.vmdk and production-data.vmdk when ever a new product needs to be supported.
So you can pick the best clone procedure for a given scenario - linked clones, full clones and sometimes simple copies may be used ...