Nutanix Protection Domain + SRM vs vSphere Replication + SRM | Nutanix Community
Skip to main content

Hello Community,

I want to configure replication and failover plans between 2 separate Nutanix clusters, both running ESXi hypervisor.

I do have Vmware SRM licenses and I am aware of these 2 possibilities :

1. Replicate VMs using Nutanix Protection Domains (either with Async DR or with Metro Availability) and configure Failover scenarios with SRM via Nutanix SRA
2. Replicate VMs using VMware vSphere Replication Appliance and configure Failover scenario with SRM

My question is which of the possibility would you recommend ? which is better in terms of performance and flexibility ?

Thanks.

Hi Ange,

When deciding a Disaster Recovery strategy you should consider RPO/RTO as primary factor.

Now following are some of things I can share related to same.

  1. Nutanix PD (while using Prism Element) or Leap or Nutanix DR (Using Prism Central) is highly recommended as the replication is native to platform and removes any external dependencies.
  2. If you have Advanced Replication or Ultimate you can also use Runbook capabilities through Prism Central which is replacement of SRM.
  3. While using SRA or vSRA with SRM you need to make sure all the compatibilities and configure all the additional components and dependencies, which makes things little complicated to setup and troubleshoot.
  4. Both the options are possible but VMware vSphere replication Appliance will not be able to replicate any volume groups if you are using them.
  5. If you are considering Metro Availability you need to consider other factors like vCenter placement, Witness.
  6. Nutanix native replication allows Async (> 60 min), near-sync (1-15 min), Sync ( 0 Min) based on your platform and available licensing tier.

Hope this information is helpful and reach out if you need more information.

F>P


Hello 

Thanks for you reply and all helpful details you provided.

I have some points to clarify and to ask :

 

If you have Advanced Replication or Ultimate you can also use Runbook capabilities through Prism Central which is replacement of SRM  ==> In my case, the Organization I work for has already made the choice of SRM; the licenses have even been purchased. So I MUST use SRM.

 

While using SRA or vSRA with SRM you need to make sure all the compatibilities and configure all the additional components and dependencies, which makes things little complicated to setup and troubleshoot  ==> As far as I know, Nutanix Protection Domains are compatible with SRM, since a SRA has been developed by Nutanix for this purpose. Is there any compatible issues I don’t know about it ? Please tell me more

 

Both the options are possible but VMware vSphere replication Appliance will not be able to replicate any volume groups if you are using them ==> Very good point. Thank you so much.

So, we could use VMware vSphere replication Appliance for regular VMs and at the same time Nutanix Protection domains should be used, especially for Volume Groups.  Right ?

 

If you are considering Metro Availability you need to consider other factors like vCenter placement, Witness ==> Very good point. Well noted

 

Another key point I want to be sure : While using Nutanix Protection Domains with VMware SRM, is it possible to have granularity at the VM level (rather than the Protection domains level) when we deal with Failover Scenarios ?  I know that VMware vSphere Replication Appliance offers  granularity at the VM level for Failover Plans. 

Could I have your advices about it ?

 

Thanks.


Hi Ange,

No Issues with that.

Make sure the SRA is compatible with the AOS and SRM version. The SRM/SRA uses Nutanix Protection domains at backend using SRA integration, which can be used for VMs as well as Volume Groups.

https://portal.nutanix.com/page/documents/details?targetId=Nutanix-SRA-for-SRM-v3_0:Nutanix-SRA-for-SRM-v3_0

The above link gives how to guide for SRA. 

Now if the required RPO is less then 60 min you will need Advanced Replication Licenses addon over Pro Tier.

For granularity you can create PD per VM or per VM group (which can be failover/failback along as consistency group)

In your case if you don’t have volume groups uses my recommendation will be using vSphere Replication with SRM. 

However I would like to iterate is first define what is required RPO and RTO so lets list down following questions.

  1. What’s is required RPO / RTO
  2. What’s available bandwidth and TTL between the sites.
  3. What Nutanix license tier is available.
  4. Are both the sites having Network (L2) extended, (are same network subnet available at both the sites.)
  5. Are Volume Groups configured for any VMs that to be replicated.
  6. Is vCenter  available at both the sites.

F>P


Hello sl.farhanparkar,

Thanks once more for these clarifications.

Below, the answers of your questions :

  1. What’s is required RPO / RTO  ==> RPO values will depend on the VMs criticality : Some VMs will required  an RPO=0, some others an RPO=1H, and so on.  The minimum required RTO is 1H for critical VMs, RTO is greater than 1H for other VMs 
  2. What’s available bandwidth and TTL between the sites  ==> Bandwidth = 100Mbps; TTL < 5ms.  
  3. What Nutanix license tier is available ==> Acropolis Ultimate
  4. Are both the sites having Network (L2) extended, (are same network subnet available at both the sites.) ==> The 2 sites are connected using an L3 link (no L2 extension) 
  5. Are Volume Groups configured for any VMs that to be replicated  ==> VGs are not planned for VMs to be replicated yet; but this could be use in a near future
  6. Is vCenter  available at both the sites ==> Yes, We do have vCenter Standard licenses for both sites as we want to use SRM. Each site will have its own vCenter. We also have VMware vSphere Enterprise plus licenses for all the nodes. 

Thanks


 

Hi Ange,

With all the information shared, and as you already have SRM you can use SRM with VRA. The link above have all the configuration steps required, go through it and post your experience, I had never tested RPO=0 with such setup, but my recommendation is use Near-sync which gives RPO = 1 min and maintain multiple snapshots in case wanted to revert back.

F>P

 

 


Thank you so much sl.farhanparkar

I will keep  you informed...