VMware Cloud Community
bsd1977
Contributor
Contributor

Placeholder VM for Resource Reservation

Hi all,

At my job, the way we generally operate is, if a project is initiated that has a requirement for virtual machines that use above a certain threshold of compute and/or memory resources, they are required to fund the purchase of a new ESXi host so as to replenish the resources their project will be occupying in the resource pool.  Typically the host is provisioned and the VM's built soon after that, no problem.  However sometimes we have projects that buy a host but don't have a need for the VM's right away....sometimes not even until 6 months to a year after the hardware was purchased.  My question is, from a capacity planning perspective, how do you look at current and available usage of your compute/memory resources for your clusters and remember to take into account, oh yeah, the Accounting team is going to come asking for their two 4 vCPU x 32 GB vRAM virtual machines in a few months.

I thought about creating a dummy VM and figuring out a way to run a process that would sustain usage of a certain amount of compute and memory resources.  That way the VM would always appear to be using those resources and my view of available resources at any given time would not be misleading.

I was wondering if any of you have run into a similar situation and what you've done to effectively manage it.

Thanks,

BD

Tags (1)
0 Kudos
1 Reply
ldesfontaines
Enthusiast
Enthusiast

Hi,

In my company, we do have a "booking" process. The only way we found to follow these requirement is to manage them out of vCenter.

In fact, we use 2 differents sources :

  1. Health check report, to follow actual resources demands
  2. An excel file, where every request that "may soon or not" be activated is listed

Using these 2 files, we made a small program that is able to tell us some key informations:

  1. How many vCPU are actually declared on the hosts and the ratio with the physical cores
  2. Same for vRAM and RAM
  3. How many vCPU will be declared if every request are activated
  4. Same for vRAM and RAM

This enables us to know when new hosts are required (we do not require project to fund them).

We also combined this analysis with vCenter Operations Manager dashboards on every clusters to know the actual resource demand (instead of the "declared" part as stated above).

Hope it helps.

Ludovic

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful.
0 Kudos