• 25 August 2016
  • 1 reply

Badge +2

1 I need to know how controller VMs work.Please guide me

2. What are the components of prism elements

3. we are doing POC on Nutanix NX-1065 model, There are 3 nodes on it and each has 10TB storage.
On One node i can see 10TB storage, How do i know its SATA or SAS(If i see from ESXi i see drive type unknown)

What i know its DAS but i just wanna confirm on each node is there any kind of RAID.

I have shutdown one node but still i can see 30 TB, hows is it possbile. it should show me 20TB.

If i shutdown second node will it still show me 30TB?

Can three CVMs be prism leaders?

If i shutdown my 3CVMs can i manage the box with cluster IP?

4. Do snapshots happen at storage level or at VM level.

5 For snapshots is there any dedicated space reserved.

6. We are using VMware hypervisor, Do CVMs and hypervisor datastores reside on SSDs.

I am learning the product may be my questions may by silly

1 reply

Userlevel 7
Badge +30
 - Thanks for reaching out! no question is too silly, and everyone has to start somewhere.

I've answered all of your questions in this post. Feel free to continue to post questions.


#1 How controller VM's work
Have you read through the Nutanix Bible yet? a lot of the nutanix internals are covered in there, including the vast majority of everything you'd ever want to know about Controller VM's

Once you read through that, let us know what remaining question or curiosity you have, and we'll be happy to help.

You may find the Architecture section best for answering these first general questions about what the Controller VM's do:

#2 Prism Element
Prism Element is simply the name for the administrative GUI for any single Nutanix Cluster.

Conversely, Prism Central is tool that can be deployed to manage many Prism Elements.

Refer to this section in the Nutanix Bible for more detail:

#3a SATA or SAS
The Nutanix software completely manages all of the storage, and presents all storage (whether it be SSD Flash or SATA Spinning Disk) as one pool.

On the backend, within our software, we manage the tiering between these devices at a granular level, so you don't have to deal with that complexity anymore.

Refer to this section in the Nutanix Bible for more detail:

#3b RAID
Nutanix does not use RAID, we use replication factor, which protect data at the block level, and distributes multiple copies of data throughout the cluster. Default is Replication Factor 2, which is two copies of all data, can you can go to replication factor 3 for 3 copies of data in larger clusters

Refer to this section in the Nutanix Bible for more detail:

#3c Shutting down nodes and Storage
Given that the data is fully protected, everything is N+1, meaning losing a node does not "delete" your storage, so if you shut down a node, you just don't automatically "lose" 10 TB

Meaning, if you had 30TB EXACTLY stored on the system, and you shut down a node, that 10 TB would still be there, since its protected on multiple nodes

Refer to this section in the Nutanix Bible for more detail:

#3d Shutting down a second node
IF you just have a single 1000 series POC box, you're likely only setup for N+1 redundancy, meaning, you can lose a node with no issues, and that would bring you to N+0.

If you shut down a second node immediately afterwards, that would bring you to N-1, which would cause the cluster to have issues.

The data path resiliency link I posted above should help with this.

#3e Prism Leaders
No, only one CVM can be Prism leader at a time.

This is covered here:

#3f Shutdown 3 CVM's
If you shut down all 3 CVM's, there will be no more Nutanix software to communicate with, so you would not be able to manage anything.

This would be like shutting down vCenter, and trying to connect to vCenter while it was shut down.

#4 Snapshots at Storage or VM level
Snapshots are at a VM level.

See this section for more detail:

#5 Snapshot reserved space in advance

#6 Where do the CVM's live
The CVM software is actually installed to the SSD(s) directly.

The hypervisor lives on the SATADOM, which is a small ~64GB internal SSD, and is separate from the SSD in the front of the server.