Help with DR setup for VMware 5.5 environment

Brett@S wrote: No, when you enable image access or fail over on the dr side, you can then use the replicated LUN. So you can present them to your esxi hosts, mount the datastore(s) contained on them and then add the VM's to the Virtual Center Inventory. (browse the datastore folders, find the vmx file and add to inventory). If you have configured the DR VC to be the same as the production one (Network names, stretched VLANS, resource pools etc) your VM will be good to go. Just remember, we're talking at a high level here, and there's more details to this than can be adequately explained over a forum. All this is easier done when RP is integrated with SRM or you use RP for VMs, but it's certainly achievable without. That being said, if you have budget and resources, it may be worth seeing if adding SRM into your design is feasible, as doing so will remove a lot of this complexity.

This makes sense and clarifies some questions I did have about this approach.

So in a DR fail over scenario, given that the DR VC will mirror the primary in configurations, the replicated LUN will not be available to the DR until I enable image access, does that mean in a controlled fail over where the primary EMC is still up we would have to "flip the switch" in Unisphere to stop making the primary SAN the master thus allowing the DR to see the replicated LUN?

Does SRM come as it's own suite or is part of another? Ideally, I would love to get SRM approved for this project, but at this point we are stretching the budget.

Once again, thanks for all the help.

joe_cole wrote: This makes sense and clarifies some questions I did have about this approach. So in a DR fail over scenario, given that the DR VC will mirror the primary in configurations, the replicated LUN will not be available to the DR until I enable image access, does that mean in a controlled fail over where the primary EMC is still up we would have to "flip the switch" in Unisphere to stop making the primary SAN the master thus allowing the DR to see the replicated LUN?

Pretty much, but you 'flip the switch" in Unsphere for Recoverpoint, not Unisphere for the Array ;

You can also test the DR copy using the various image access modes available for testing without affecting access to the Production data.

SRM licensing is available in a few options, including packs of x amount of vm's. Best to check with your VMware rep for pricing, but it's not overly expensive IMO, and provides a great way to wrap all this up and tie it all together.

The other option would be talk to your EMC rep about swapping in a Recoverpoint for VMs license, instead of the Remote Protection Suite.

This takes the need for SRM out, and is a pure EMC solution that will do what you want nicely.

Brett@S wrote: joe_cole wrote: This makes sense and clarifies some questions I did have about this approach. So in a DR fail over scenario, given that the DR VC will mirror the primary in configurations, the replicated LUN will not be available to the DR until I enable image access, does that mean in a controlled fail over where the primary EMC is still up we would have to "flip the switch" in Unisphere to stop making the primary SAN the master thus allowing the DR to see the replicated LUN? Pretty much, but you 'flip the switch" in Unsphere for Recoverpoint, not Unisphere for the Array ; You can also test the DR copy using the various image access modes available for testing without affecting access to the Production data. SRM licensing is available in a few options, including packs of x amount of vm's. Best to check with your VMware rep for pricing, but it's not overly expensive IMO, and provides a great way to wrap all this up and tie it all together. The other option would be talk to your EMC rep about swapping in a Recoverpoint for VMs license, instead of the Remote Protection Suite. This takes the need for SRM out, and is a pure EMC solution that will do what you want nicely.

Happy New Year Brett

It appears I might be able to present my case to the project manager for an additional add-on for this project. This could either be SRM or RecoverPoint for VMs.

I'm kinda leaning towards RP for VMs since it will be a centralized EMC solution, however not sure if their is a limit of VMs or license based on VM count.  Ideally, would like to go with a solution that more seamless where less manual intervention is involved.

Any thoughts?

Thanks

HNY to you too Joe

Good you have some scope to look at some options.

I think the minimum RP for VM pack is 15, and I'm not sure of the licensing increments.

Having it centralized is a good point, as is the integration (the splitters are installed directly via VIB's).

A very appealing feature is that you can do VM level recovery as well as LUN/Datastore level. You get pretty much the same GUI orchestrations/testing functions etc as you do with SRM.

If I were doing a purchase and was looking solely to protect VMware VM's, RP for VM's would be my choice.

Drill your EMC rep to do you a good deal, as the Remote Protection pack is an expensive option (particularly when you consider maintenance) that you have paid for. Would be nice of them to work something out for a changeover.

Brett@S wrote: HNY to you too Joe Good you have some scope to look at some options. I think the minimum RP for VM pack is 15, and I'm not sure of the licensing increments. Having it centralized is a good point, as is the integration (the splitters are installed directly via VIB's). A very appealing feature is that you can do VM level recovery as well as LUN/Datastore level. You get pretty much the same GUI orchestrations/testing functions etc as you do with SRM. If I were doing a purchase and was looking solely to protect VMware VM's, RP for VM's would be my choice. Drill your EMC rep to do you a good deal, as the Remote Protection pack is an expensive option (particularly when you consider maintenance) that you have paid for. Would be nice of them to work something out for a changeover.

At this point, would you say having the current version of RP worth keeping? From my understanding, we aren't looking to replicate any other type of LUNs except for VMWare LUNs.

No, you wouldn't have both RP and RP for VM's (if that's what you mean). You'll only have one or the other. (at least in the context of this discussion)

Even more incentive to use RP for VM's, it's now Free to download/use, just pay for support. Very progressive model.

First present of 15: Free to use Recoverpoint for VMs! - Virtual Geek

Will be the same...the RPA's will still split and handle the I/O replication through their WAN virtual networks. 

-Keith

Hey Brett,

So here's the current status. Going to go with SRM and leverage Recoverpoint for the replication.

Question I have is do I have to setup a WAN interface on the vnx5400 for replication or is this done via SRM? If so, does that require the WAN connection to be on the ESX server it self?

Hi Keith,

Thanks for the info, how would it be if were to leverage the vRPA instead?

Hi Joe,

If you are going with SRM and native RecoverPoint (physical appliances leveraging the RP SRA), then the replication mechanism will be done through the RPA's and RecoverPoint versus on the VNX itself. Each RPA will require a LAN and WAN connection and it is the WAN connection that will replicate the data between peer RPA's on the DR site. SRM will interact with RecoverPoint through the SRA by giving it the site mgmt. address of each RP cluster (Source and DR) and facilitate the failover/test mechanisms when invoked through SRM. There won't be any required WAN setup on the VNX or ESX servers using SRM with native RecoverPoint appliances.

HTH,

-Keith

Currently allocating a 10GB interface which will be trunkd for basic VM networks. Would allocating additional vlans on this interface for the vRPA be enough?

From a functionality perspective, different vlans should work as long as they are segmented per the requirements. From a performance perspective, you might want physical isolation on the interfaces, especially since you need isolation for the iSCSI and LAN/WAN networks. Here are the requirements for isolation in this guide on page 39/40: https://support.emc.com/docu9565_RecoverPoint_Installation_and_Deployment_Guide.pdf?

The current setup already has iscsi separated on its own 10GB interface via 2x multipaths so this may work.

I would hate to ask for additional 10GB card for 8-9 servers... ahhhhhhh