These days most of my replication is handled at the VM-level by software design for virtualization. While that is the case for most of my evironment, I still have a few non-virtualized workloads that run on shared storage that need to be replicated in the event of a disaster at my primary location. This process has never been too complex from my days of working with NetApp and now as I continue exploring the Tegile I’m happy to say that it’s just as easy through the GUI.
Documenting this process for my non-virtual workloads would be a little difficult so I’ve decided to document this process using an NFS datastore containing a few virtual machines. The first half of this guide is setting up the replication relationship and replicating the data. The second-half is the process to actually restore that data and make it usable at your DR site.
1. Login to the web interface of the Tegile that is the replication source
2. Click on “Settings” then “App-Aware”
3. Click on “Zebi Replication” on the left column
4. Under the tab “Replication Target” click the “Add” button (This is adding the DR Tegile as the target array)
5. Enter the name or IP of the array (the shared Management IP address) and the username/password (Optionally you can specify a port range for replication which we won’t be doing for this documentation) and click “Add”
6. Once it has been successfully added it will appear in the “Replication Target” list
7. Login to the web interface of the DR target Tegile, click on “Settings” then “App-Aware”, choose “Zebi Replication” on the left column and then click on “Replication Source” tab. You should see your other array listed here (The IP address will be the “management” IPs of each controller, not the shared management IP for both arrays)
8. Back on the Primary Tegile (Replication source) click on “Data”
9. Click on the disk pool then then project that will be replicated
10. For this documentation I’ve created a Project named “NFS_Replication” with a volume named DR_Windows with 4 VMs inside. Click on the project that will be replicated and click on the “Edit” button
11. Click on “Replication” on the left column
12. Click the “Add Replication” button
a. Select the Target System and click “Next”
b. Select the “Target Pool” and enter a name for the “Replication Project”. Click “Next”
c. Choose what options are required and which volumes will be replicated (This test only has one volume, DR_Windows, but you can include or exclude any volumes that exist in this project. We’ll choose quiesce which will perform a VMware snapshot to put the OS in a consistent state. Click “Next”
d. Choose your schedule (manual or automatic), frequency, and how many additional snapshots (restore points) will be saved on the target array. For this example we’ll do daily replication that happens at 10:49 am and we’ll keep 14 snapshots. Click “Finish”
e. Once it’s all setup, you’ll see your target array, the target pool, and the target project
13. I have 4 VMs in that datastore (DR-Test01-04). Once the time hits, we can see that snapshots are taken, then removed, for each of the VMs in that datastore.
14. On the DR target array, we can see we now have snapshots available for this project. (The reason there are 2 is because I initiated a manual replication sync for testing first)
a. To manually kick off a replica snapshot, on the source array, find the project, click on “Replication” and then click the “Play” button that says “Replicate”
That is how simple it is to setup replication. Now let’s imagine we need to spin up those replicated VMs in this volume. Here is how we do that.
1. On the DR target array, click on Data, select the pool, then click on “Replica (1)” to view the replica project
2. Click the “Edit” button for the NFS volume
3. Click on “Snapshots” and find the snapshot you want to bring live (We’ll choose the latest version). Click the “Clone” button
a. Cloning the snapshot will allow us to create a new project and NFS Volume from this snapshot and spin up these VMs in DR. By doing a clone, we’re able to continue to replicate data in the event you are testing replication instead of having an actual DR event.
4. Enter a name for the new Project (DR_NFS_Replication for this writing) and a name for the mount point (/export/DR_NFS_Replication for this writing) and click “Clone”
5. If successful, you’ll receive this message about the new project being created. Click “OK”
6. Close the window for “Share Configuration” and click on “Local (1)” under “Projects”
7. Click on the “DR_NFS_Replication” project then view the Mountpoint of the Share (/export/DR_NFS_Replication/DR_Windows). Note the “c” before the share name which denotes it was a clone from another projects
8. Click the “Edit” button for the project and then click on “Sharing”
9. This is where you will add the IP addresses or range of IPs that need read/write and root access to the shares in this project. The IP addresses/ranges will carry over from the source array. Our IP range is the same in DR as our lab so we’ll leave this alone.
10. Connect to your DR vCenter server or ESXi hosts. Click on the host, then “Configuration”, then “Storage”
11. Click “Add Storage” towards the top right
12. Choose “Network File System” and click “Next”
13. Enter the NFS IP address of the DR Tegile, enter the folder path (/export/DR_NFS_Replication/DR_Windows) and then enter the name of the Datastore (DR_Windows). Click “Next”
14. Review the summary info and click “Finish”
15. Repeat for each host that needs access to this datastore. Afterwards, right click the datastore and click “Browse Datastore”
16. Inside you’ll see the 4 VMs we that were located in here before. Open each folder, right click the VM name.vmx file and choose “Add to inventory”
17. Enter the name and location for the VM and click “Next”
18. Choose the Cluster or host and click “Next”
19. Review the settings and click “Finish.” Repeat for each VM that needs to be added.
20. Power on all the VMs and now you can run any validation tests or bring these VMs live in a DR event
Obviously, the process of mounting the datastore in your DR vCenter Server and re-adding the VMs one by one would be time consuming and tedious. When developing your DR plans, having this process scripted (easy enough in something like PowerCLI) ahead of time on the vCenter side of things would ease that burden. From the standpoint of the Tegile, this process is fairly intuitive and simple to setup. One of the things I love is that by default the data you are bringing live on the DR site is a clone and replication continues running without being affected.