Skip to main content
Solved

How to Size a new Nutanix cluster from VMWare to Nutanix

  • November 19, 2022
  • 8 replies
  • 1453 views

Hello All,

 

I'm working with a customer on a new solution design

the customer has a VMware workload and we plan to put AHV,

I want you to advise me on how I calculate the storage requirement, Let’s say the customer workload is about 170 TB how do i size the cluster storage

 

Also, I have some questions related to storage sizes

if I have 3 nodes cluster

each node with 10 disks each disk 4 TB (numbers for examples only)

that means total raw storage is 120 TB

we will go with RF2 which means 1 node is out of our usable capacity calculation

 

So what is the usable capacity here after the 1 node for failure and the Raid used in the cluster creation, also what else should be conducted from the storage in storage planning?

Best answer by qamarabbas

Hi, 

Normally Considering RF#2 means if you have 120TB RAW Capacity then considering RF#2 logical Capacity is 60TB.  You can use upto 60TB but considering n-1 you have to reconsider 1-node storage free all the time. 

This topic has been closed for replies.

8 replies

Moustafa Hindawi
Forum|alt.badge.img+6

Hello @Teko Yasin 

First, to calculate the usable capacity you need it's better to do it via Nutanix Sizer, and you have to provide workloads usable storage, however to convert RAW storage to usable storage with RF2 HDD is not enough, in Nutanix SSD contribute in the usable capacity. Calc all (HDD + SSD) x 0.82 / 2 = usable storage.

Second, you have to check hardware compatibility.

Third, the usable capacity after Node failure is very simple, after doing the above calculation, if they are 3 node sub. 33%.

 Forth, Nutanix doesn't use RAID, Nutanix uses Replication Factor+Data locality. which means in RF2 one full copy of the vm located where the vm reside, and the second copy chunks to 1-4 mb and distributed on the entire nodes in the cluster.

 

looking forward to your feedback or comments


  • Author
  • Adventurer
  • 3 replies
  • November 19, 2022

Hi Moustafa Hindawi

thank you for your reply, can you explain what is the sum (calc all (HDD + SSD) x 0.82 / 2 = usable storage) based on?

and should I remove 1 node after this sum to get the usable capacity?

ex: 

3 nodes cluster with 406 TB (SSD+HDD)

The sum will equal:166.46

usable will be: 110 TB?

 

406 TB raw storage provides only 110 TB usable !?


Moustafa Hindawi
Forum|alt.badge.img+6

Hello @Teko Yasin 

If all disk (SSD+HDD) RAW = 406TB then we will sub. slack and overhear 18% = 333 then divided on the Replication factor 2 (Like if you do for RAID1 in traditional solution) = 167 TB « that’s the usable storage with data redundancy 2 copies, in node failure scenario will be = 110 TB  your data still available and no data lose. and you can add from 10-30% data saving (Compression and Dedup) will be for ex. 132TB N-1 or without Node failure will be 200TB

On the other hand (vSAN) (in a simple way) the same calc BUT you will remove the SSDs size from the calc because in vSAN SSD doesn’t contribute in the data store, and increase the slack and overhead to 25-30% instead of 18% 


  • Author
  • Adventurer
  • 3 replies
  • November 19, 2022

what do you mean by we will sub. slack and overhear 18%?

 


Moustafa Hindawi
Forum|alt.badge.img+6

Yes subtract 18% from the total RAW capacity then divide on the number of replication factor 2.

I hope that’s answer your question, and if you need anything please let me know


  • Author
  • Adventurer
  • 3 replies
  • November 19, 2022

I got the meaning but I’m asking what is the 18% represent.


Moustafa Hindawi
Forum|alt.badge.img+6

Disk balancing, garbage management, write buffer.


qamarabbas
Forum|alt.badge.img+4
  • Vanguard
  • 117 replies
  • Answer
  • November 20, 2022

Hi, 

Normally Considering RF#2 means if you have 120TB RAW Capacity then considering RF#2 logical Capacity is 60TB.  You can use upto 60TB but considering n-1 you have to reconsider 1-node storage free all the time.