We have 20TB Nutanix cluster (3 nodes of ESXi). Currently Replication Factory is set to 2 (seems its default). I want to leverage complete 20TB and we don't need any redundancy. Is this possible?
With RF2, it seems we are able to use only 10TB and not 20TB. We don't keep any important data and can tolarate failures but idea is to use complete 20TB instead of 10TB. Any help?
Using: Nutanix Version 4.1.4
What is your use case? What do you have on there that you can lose that much data if a drive fails?
We create N numbers of application VM's which are fine if goes away due to disk failure. We can recreate application vm's again from code as those are mode of dev/staging vm's. E.g currently i create 200 VM's in 10TB, then i will be able to crete 400VM's if we don't do redudancy.
More you can help how to achive it.
I'm curious, how large are these VM's on CPU, Memory, and Storage? Just because you have the storage to provision these, I'd imagine you'd run out of CPU and Memory at some point?
Also, I'd recommend you run AHV if you aren't already.
Not only will this help you optimize cost, but it also makes cloning and recreating VM's REALLY easy, with VERY space efficient clones. Seriously, this works really well.
You can do something similar on VMW and Hyper-V with VMW VAAI and MS ODX, respectively, but AHV clones are by far the most efficient and easiest to use.
I'd also recommend that you run in-line compression in this setup.
Using both space efficient clones in AHV, and in-line compression will help you regain a LOT of space with a dev/test setup.
Basically, if you have the compute and memory for 400 VM's, I'd rather try to help you find a way to fit them in with RF2, so that you dont have to go re-provision many VM's after some sort of failure, and have to spend your time working on failure recovery, rather than just replacing a disk and going on with your day
As a opposite, Even it`s a rare circumstance with RF1, I also have a question about RF4 or greater RF. I thinks there indeed are some requirement to use larger RF to achieve strong local NODE/disk protection without DR in a large environment. because RF3 can only tolerate 2 NODE failure
As I replied on a separate thread, RF4 would be neat, though we have yet to find a customer where this configuration was required.