vCSA in a homelab?

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Cape

Member
Oct 28, 2015
36
6
8
Hi!
I'm about to add a new esxi node, and figured I'd probably want vCenter as well. However, the sizing seems pretty brutal for a homelab, with the "tiny" option claiming 2 vCPU/10 GB mem/300 GB disk? I obviously don't have the budget/space some of you lucky people have (one day!), but this seems a bit prohibitive unless you are putting up a rather big environment.
For reference, my old node is a X9SPV-LN4F-3QE (4/8 core, 12 GB mem), and the new one is a Xeon D-1528 (6/12 core, 32 GB mem). So a 10 GB footprint for vCSA would be quite a big chunk of my resources... How do you manage this? Or am I simply not playing big enough?

My motivation for vCSA would be partly to learn about the HA/vMotion bits (in practice, letting the pfSense VM migrate between nodes), and part in having a single point of administration for virtualization, vs logging into specific nodes for specific VMs.
 

msg7086

Active Member
May 2, 2017
423
148
43
36
vCenter is kinda enterprise facing product. If you can afford $2000 vCenter foundation, or $7500 standard, you probably won't be worrying about 300GB disk space.

For pfSense VM, failover would be much more affordable than a vCenter. IIRC, vCenter clustered host HA requires very large amount of resources just to replicate the memory space. While pfSense failover is kinda free. (Correct me if I'm wrong.)
 

msg7086

Active Member
May 2, 2017
423
148
43
36
With that said, I checked one of my lab instance. It's only 50GB after a while of use. Don't use thick disk provision. Memory wise it actually uses less than 1GB actively, meaning that you can probably reduce the memory settings to maybe 4GB. (I never tried, you are on your own.)
 
  • Like
Reactions: Iaroslav

Cape

Member
Oct 28, 2015
36
6
8
Yeah, the 300 GB disk wasn't the main issue, it seems over the top to keep track of a handfull of hosts, just like the 10 GB of memory. I guess it is partly because as you mention, this is not optimized for the lab, so doing resource optimization might not be something their customers really case about/want to pay for.
I installed it, and after 12 hours it actually uses the full 10 GB of memory. Probably it just allocates it right away to have it for caches or something.

So, seems I won't be running vCenter (disk is ok, but I can't really motivate using over 20% of my memory for it). What are people doing in their home labs instead? Just manually managing individual hosts?
 

msg7086

Active Member
May 2, 2017
423
148
43
36
You can always try to downscale the spec and see if it boots up correctly. On the other hand, don't lock it in the physical memory so host can swap its unused memory out to disk (since most of the time it'll just run idle).

I personally run individual proxmox boxes and I don't feel the need to move VMs around, etc.
 

DRAGONKZ

Member
Apr 9, 2018
87
10
8
41
The only way you can normally shrink the tiny resource requirements is to perform the install, stop/disable services for things you won’t be using, then adjust RAM/CPU acccordingly.

There used to be a few guides with people explaining how to do this in the 6.0-6.5 days, perhaps search for them on google.

vCenter is meant for corporate use, where a 16GB requirement for a small VCSA would be a drop in the ocean in terms of resources available to use.

I’m personally creeping up to the 1TB RAM mark in my 4 node lab, so RAM usage is the least of my concerns (disk is my concern, haven’t got my vSAN running yet so running low!)
 

Dawg10

Associate
Dec 24, 2016
220
114
43
How do you manage this? Or am I simply not playing big enough?
It sounds like you want to play big with a resource constrained system; you will be disappointed with the results. Stuff will work but it'll be a pain in the ass; the "Tiny" configuration is designed for this. My VMware/ VMUG virtualization journey is only 2 years old; I keep wanting to do more and have to up the hardware ante in order to do so (HA requires 3 hosts, so I now have 3x identical R320's). One of the primary upgrades was to provision VCSA with enough resources to handle the tasks being asked of it; at the moment that looks like 2vcpu & 24GB ram on a 12core 256GB host with only one other ESXi host online, another 12core 96GB system.

When it comes to a satisfying experience, size matters.
 

fishtacos

New Member
Jun 8, 2017
23
13
3
I keep wanting to do more and have to up the hardware ante in order to do so (HA requires 3 hosts, so I now have 3x identical R320's).
HA requires 2 hosts - you might be thinking of vSAN and HA combined together, which ups the requirements to vSAN's minimums, which is 3 (or 2 with a witness host)
 

Dawg10

Associate
Dec 24, 2016
220
114
43
HA requires 2 hosts - you might be thinking of vSAN and HA combined together, which ups the requirements to vSAN's minimums, which is 3 (or 2 with a witness host)
But you can do sooo much more re: DRS/ HA with 3 hosts... In the current config, the primary server is in a cluster with all 3 R320's; the primary is loaded up to the point it takes 2 R320's to offload the VMs. Playing with DPM and Orchestrator for shits and giggles.

My point was: with ESXi you have to pay to play. Pay more = play more. I wish I could afford a pair of R740s.
 

Cape

Member
Oct 28, 2015
36
6
8
Thanks for all the info/reference values! I'm _very_ space constrained, unfortunately, so expanding is physically impossible :/
I'll do some thinking on how to do the best with what I have, possibly asking for some feedback in a new thread :)
 

jerbkazzaz

New Member
Feb 12, 2019
2
0
1
Hi!
I'm about to add a new esxi node, and figured I'd probably want vCenter as well. However, the sizing seems pretty brutal for a homelab, with the "tiny" option claiming 2 vCPU/10 GB mem/300 GB disk? I obviously don't have the budget/space some of you lucky people have (one day!), but this seems a bit prohibitive unless you are putting up a rather big environment.
For reference, my old node is a X9SPV-LN4F-3QE (4/8 core, 12 GB mem), and the new one is a Xeon D-1528 (6/12 core, 32 GB mem). So a 10 GB footprint for vCSA would be quite a big chunk of my resources... How do you manage this? Or am I simply not playing big enough?

My motivation for vCSA would be partly to learn about the HA/vMotion bits (in practice, letting the pfSense VM migrate between nodes), and part in having a single point of administration for virtualization, vs logging into specific nodes for specific VMs.
I have a vCenter appliance in my homelab and keep the Tiny sizing. When I was more resource constrained, I lowered the RAM down to 6 or 8GBs and didn't see much of a difference. It was slower overall - but don't recall any showstopping issues.

Also, are you thin-provisioning your disk on the VCSA VM? I'd say there's no advantage for a homelabber to chew up the full 300GBs on your storage. The performance difference between thin vs thick is super negligible in this case. I'm sure you can find better use for all that disk instead of it being taken up by an empty VMware database. :)
 

m3xiz

New Member
Jan 21, 2019
11
1
3
I wouldn't try to play with vCenter/vCA just for pfSense. This is over-killed as pfSense has the HA built in. I am 99.99% sure that doing HA using the pfSense feature will be more reliable than playing with vCenter.
 

dswartz

Active Member
Jul 14, 2011
610
79
28
Also, the cost is somewhat overstated. If you spend about $200/yr for a VMUG membership, you get a full enterprise license for vsphere, including vcenter server appliance, etc...
 
  • Like
Reactions: CreoleLakerFan