Build 3 DELL node for test ON

We would like to test and evaluate opennebula on the following nodes

3 nodes
DELL R630
Dual socket 10C/20T
256GB RAM
2 x SSD Raid 1 > operating system
6 x SSD direct > storage
10Gbit network

The goal is to create a kvm cluster, managed with opennebula, with HA for VM and ON Cloud Frontend

We are starting to delve into the documentation, but we have several basic doubts :slight_smile:

  1. we need to build first node with KVM onboard and then create VM for Opennebula Cloud Frontend?
  2. the other 2 node can be auto deployed from ON Cloud Frontend as ON Edge ?
  3. building of storage for HA can be choice/managed from Cloud Frontend or must done direct on single node? ON have integrated solutions or we need, for example, to setting CEPH or similar?
  4. if we use Edge cluster the ON hybrid cloud 3-tier replica what’s done exactly?
  5. we can use 3 nodes for: VM ON Cloud Frontend and VMs, all under full HA, container and object storage services?

We speak always about Enterprise Edition

Thanks for any help, support, sugges :slight_smile:

Dear Simon,

sorry, I don’t have a solution for you but a similar problem as we plan a small 5-8 node hci kvm cluster and were wondering if Open Nebula could do the job maybe better than Proxmox or other solutions.

Have you made any progress with your tests? Have taken a look in OneStor from Open Nubula (OneStor Datastore — OpenNebula 6.0.4 documentation )?

Thanks & best wishes,
Ralf :smiley:

  1. we need to build first node with KVM onboard and then create VM for Opennebula Cloud Frontend?

You can choose to deploy the frontend in a VM or directly in the node hardware. Just note that the VM shouldn’t be managed at all by the frontend (itself) because chicken and egg. Note that you can have a node to work both as a KVM host and the frontend.

  1. the other 2 node can be auto deployed from ON Cloud Frontend as ON Edge ?

Yes, you can do such thing with public cloud providers and on-premises hardware. Take a look here.

  1. building of storage for HA can be choice/managed from Cloud Frontend or must done direct on single node? ON have integrated solutions or we need, for example, to setting CEPH or similar?

For HA we don’t provide a custom solution, just a requirement of the storage view to be the same across all of the frontends. You can do so, for example, with NFS mounts or cephfs. More details here

  1. if we use Edge cluster the ON hybrid cloud 3-tier replica what’s done exactly?

The exact details are here. Basically, there will be 3 Frontends running an OpenNebula instance. One of the will be the leader, that will issue database query commands to the followers to keep the state in sync. They sent heartbeats to each other to keep an eye out for their presence. One of them no longer beats, the rest will decide which one will be the new leader.

  1. we can use 3 nodes for: VM ON Cloud Frontend and VMs, all under full HA, container and object storage services?

Yes. Note that there are two HA, frontend HA and VM/Hypervisor HA.