Remark: All of the information are based on my installation experience for VxRail Appliance and Nutanix Metro Availablity. The following information is for your reference. Please let I know once if it has any mistake.
High Level Diagram of VMware vSAN Stretched Cluster on VxRAIL
High Level Diagram of Nutanix Metro Availability
Management
vSAN Stretched Cluster on VxRAIL – Central Management Console (vCenter Server).
Nutanix Metro Availability – Different Management Consoles (vCenter Server + Prism Central Home Dashboard).
VM Workloads
vSAN Stretched Cluster on VxRAIL
- Workloads have access to the combined resources between Site 1 and 2.
- All the resources are treated as one single cluster.
Nutanix Metro Availability
- Workloads need to be associated with a storage container.
- An active container on one site can have a stand-by (passive) container on another site.
Design and Performance
vSAN Stretched Cluster on VxRAIL
- Kernel-level integration with VMware not only enables higher performance.
Nutanix Metro Availability
- Controller virtual appliance (CVM) on every node controls I/O – can be its own “noisy neighbor”.
Day-2 Operations – Planned Maintenance
vSAN Stretched Cluster on VxRAIL
Failover (Site 1 to Site 2)
- Just move all VMs which are running at Site 1 to Site 2 by vMotion.
Failback (Site 2 to Site 1)
- Just move all VMs which are running at Site 2 to Site 1 by vMotion.
Nutanix Metro Availability
Failover (Site 1 to Site 2)
We must perform the following operations.
1. Update Affinity Rules for all VMs (Site 1).
2. vMotion VMs to Site 2.
3. Promote the Metro protection domain (PD) in Site 2.
4. Disable the PD in Site 1.
5. Re-enable the PD in Site 2.
Failback (Site 2 to Site 1)
1. Update Affinity Rules for all VMs (Site 2).
2. vMotion VMs to Site 1.
3. Promote the Metro protection domain (PD) in Site 1.
4. Disable the PD in Site 2.
5. Re-enable the PD in Site 1.
Data Protection
vSAN Stretched Cluster on VxRAIL
- Storage policies defined for each VM (virtual disk level).
- In vSAN Stretched Cluster, we can define storage polices to make a full copy (minimum) of each VM that store on local site, remote site, or both local and remote site.
NOTE: The number of virtual machine’s full copy that is depended on the settings of storage polices.
Nutanix Metro Availability
- Nutanix storage volume is mounted within vSphere. It only replicate the data of active controller to standby controller.
- Nutanix only solves traditional SAN complexity issue but not design for VMs.
VMware Integration
vSAN Stretched Cluster on VxRAIL – Deep VMware Stack Integration.
Nutanix Metro Availability – Need to wait for certification at VMware Compatibility Guide for every Nutanix product and new version.
Licensing
vSAN Stretched Cluster – Enterprise Edition.
Nutanix Metro Availability – Requires Ultimate edition software.
Deduplication and Compression
vSAN Stretched Cluster – Supports on All-Flash mode only.
Nutanix Metro Availability – Includes of Hybrid and All-Flash.
Software Packages Upgrade
vSAN Stretched Cluster
- All-in-one software package upgrade through VxRail Manager, it includes of Integrated Software, vCenter Server Appliance, vSphere, vSAN, BIOS, NIC driver & firmware etc.
- The software upgrade can be executed automatically
Nutanix Metro Availability
- Requires to upgrade Nutanix Prism software and VMware software separately.
VxRail is just a better solution… rock solid, more integrated. If you want HCI, and VMware, go VxRail.
LikeLiked by 1 person
You must not have tested the Nutanix failover as you list a process that doesn’t use the witness at all. The witness is there to completely automate the failover and failback process for VMs, just as it is in the VXrail setup. Without the witnesses neither solution works. In a Site 1 failure with Nutanix Metro, all the PD actions are done automatically and then Vmware HA restarts the VMs at the secondary site. No admin intervention required. Just wondering why you don’t list that , as that’s the way it works.
LikeLike
For Unplanned Maintenance, if Site 1 failure with Nutanix Metro, all the PD actions are done automatically and then VMware HA restarts the VMs at the secondary site. But for Planned Maintenance, both are different.
LikeLike
Victor, have you done a Nutanix Metro Availability implementation? just curious.
I do suggest watching these videos made in 2015, yes! 2015.
http://www.joshodgers.com/2016/06/15/metro-availability-witness-failure-scenario-9-network-partition-site-failure/
Cheers
LikeLike
Please check this guide (page 4) which is from Nutanix. For planned failover, we need to do some manual tasks, please correct me if my understanding is incorrect.
https://d6g9c8a9.stackpathcdn.com/wp-content/uploads/2016/08/vSphere-Metro-Availability-Work-Instructions.pdf
LikeLike