vVols – FC and iSCSI together

Let’s continue our previous discussion around iSCSI by looking at accessing vVols with FC and iSCSI in the same datacenter. Beginning with the Embedded Dell EMC VASA Provider version 3.0 (aka eVASA), we support the use of iSCSI with the Protocol Endpoint. (Note it is only the eVASA version that supports iSCSI. You can’t use the external VASA Provider 2.0 (version 9.2) with iSCSI, only FC.) For this post I am going to assume two things about the vVol environment besides that FC and iSCSI are setup: First, that the 2 eVASA Providers are registered in the vCenter; And second that there is a storage container of some size available on the array. If you need to do those tasks and are unsure how, I have documentation you can find and follow in the library. Onto the provisioning then.

Environment

Again, we’ll start with a look at the environment. I am using the 2 hosts from the previous post, dsib0184 and dsib0186. These ESXi hosts have both FC and iSCSI configured, though I am only going to use one protocol for each host. The vCenter has a single datacenter, London, with 2 clusters, London_Cluster_1 and London_Cluster_2. Each has a single ESXi host which isn’t exactly a cluster but working with what I got.

Neither of these hosts currently has a Protocol Endpoint (PE) provisioned and no vVol datastores. Therefore, we’ll start with that. I’ll provision a PE from PowerMax array 197600450 as follows:

  • dsib0184.lss.emc.com – FC PE
  • dsib0186.lss.emc.com – iSCSI PE

vVol Dashboard – Provision PE

Using Unisphere for PowerMax, navigate to the vVol Dashboard (Storage -> vVol Dashboard) and on the right-hand side under Actions, select PROVISION PROTOCOL ENDPOINT TO HOST. I’ll show the wizard for both protocols, but only the main dashboard once, as it is step 1 for both:

Fibre Channel PE

Start by select the FC host initiator group in step 2. You’ll notice I actually have 3 protocols there – FC, iSCSI, and FC-NVMe (dsib0184_fn_ig). We have the important information on the right-hand side about unique PEs. Basically it says you can’t have more than 1 PE presented to a single ESXi from an array. You could have multiple PEs if you have more than one array doing vVols, just not more than one from each array. Also, though not spelled out here, you can’t present the same PE with two different protocols like FC and iSCSI. VMware does not permit this.

In step 3, either select an existing port group or create a new one as I have done here:

Finally, in step 4 you can change the masking view and storage group name, if you wish, to something more descriptive. For example I always include “pe” in my autoprovisioning names.

A quick rescan of the adapters will show the 3.75MB device. Remember, that though there is a Protocol Endpoints screen in vCenter (I’ll show this further down), you won’t see the device there yet since we have not created a vVol datastore.

Now let’s do iSCSI.

iSCSI PE

Skipping that first step above, select the iSCSI initiator group for the other host, dsib0186:

I’m creating a new port group for my iSCSI masking view:

And doing the same type of rename:

Rescan the iSCSI software adapter and voila:

Create vVol Datastore

Now that each of our ESXi hosts has a PE, we can create a vVol datastore. Since I have 2 different clusters, but intend on sharing the vVol datastore across both the FC and iSCSI host, I am going to create the vVol datastore at the Datacenter level. This will give me the option of creating/mounting the datastore to both hosts at once. If you create the datastore at one or the other ESXi hosts, you will have to mount to the remaining one after.

Step 1 – At the datacenter level select New Datastore:

In step 2 choose the radio button for the vVol selection:

Step 3 directs you to choose the desired storage container on the array. I have two on my array:

At step 4 we are given the option of selecting both hosts for accessibility since we are creating at the datacenter level.

And in step 5 review:

Once you finish, VMware will issue 2 create datastore commands, one for each server. It issue the commands sequentially as the first must complete (or fail) before it issues the second. Below you can see both complete very quickly:

Now that the datastore is available here,

we can look at the Protocol Endpoints screen mentioned previously. Here you will see the FC PE for ESXi server dsib0184 and the iSCSI PE for ESXi server dsib0186.

So we know the protocols can co-exist, therefore let’s move a VM between the hosts as if we were doing a migration.

vMotion FC to iSCSI

Now that I have a datastore on each host, one with a FC PE and one with an iSCSI PE I can demonstrate that you can vMotion a VM between the ESXi hosts despite the vVol datastore being accessed through different protocols. This one I’ll do through a looped GIF as screenshots won’t convey it the same.

Conclusion

I expect most customers will use one protocol or the other, but hopefully this provides some comfort for those who will use both, or perhaps migrate between them.

Advertisement

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Website Powered by WordPress.com.

Up ↑

%d bloggers like this: