In this NetApp training tutorial, we will discuss the NetApp SyncMirror. NetApp SyncMirror is an optional feature which provides redundancy for our disk shelves. Scroll down for the video and also text tutorial.
NetApp SyncMirror Video Tutorial
I saw the writing on the wall about new technologies and wanted to transition to a new role as a storage / HCI cloud architect.
I used Flackbox.com classes to sync me up with ONTAP and passed my certs two days after completing your training, it took about six business days in total.
I got the job I wanted with CGI and it’s a great company to work for.
There are multiple levels of redundancy for our disks and shelves within a NetApp ONTAP storage system. RAID-DP provides redundancy in case we lose up to two disks in a RAID Group in an aggregate. Multi-Path High Availability (MPHA) provides redundancy for our connectivity from the controllers to the disk shelves. We have two SAS connections going down to our disk shelves, so if we lose any of the connections or any of the shelves, we can still get to the other functioning shelves.
So RAID-DP gives us redundancy for a couple of disks failing, and MPHA gives us redundancy for our connections, but what if an entire disk shelf fails? We want to have redundancy for that as well. That's where NetApp SyncMirror comes in.
When SyncMirror is enabled on an aggregate, the aggregate is made up of two sets of disks which are mirrored. As a result, whenever we write to that aggregate, we write to both sets of disks at the same time. For this functionality, you're going to need twice as many disks to create your SyncMirror aggregate.
The benefit we get from this is we get data redundancy. If we lose a disk shelf, we don't lose our data. The drawback is it's going to be more expensive because you're going to need twice as many disks.
You don't necessarily need to do this at a system-wide level. You can also do it for targeted aggregates if you prefer, you don't have to do it for all of your aggregates.
Along with controller High Availability, SyncMirror is the enabling technology for MetroCluster which we'll cover in a later tutorial.
The two different mirrored sets of disks that make up a SyncMirror aggregate are called Plexes. The two plexes are on different shelves, connected to the system via separate cables because the point of SyncMirror is redundancy for our shelves. There would be little point in running SyncMirror if the disks were on the same shelf. A normal, non-SyncMirrored aggregate contains only one plex.
We also have pools. To configure SyncMirror, the disks must be divided into two pools by the administrator. The two pools will be Pool 0 and Pool 1. Disks for the first plex in a SyncMirror aggregate are assigned from Pool 0 and disks for the second plex are assigned from Pool 1. There's a one to one relationship between plexes and pools.
When disk auto assignment is turned on (which is the default), all disks are assigned to Pool 0. To configure SyncMirror, you're going to have to disable auto assignment and then manually assign half the disks to Pool 0 and the other half to Pool 1.
NetApp SyncMirror Configuration
The diagram below shows a new system with all of the default settings. All of the disks are in Pool 0. We need to have an Aggregate0 for our system information, which is how it comes configured straight out of the factory. Aggregate0 is made up of three disks configured with RAID-DP.
As an administrator, I would then need to create my first aggregate for user data. I've called it Aggregate 1 here. Again, by default, all the disks will come from Pool 0. This is not a mirrored aggregate. In our example, it's made up of a couple of different RAID groups. Notice all my spare disks are in Pool 0.
Now I want to convert Aggregate 1 into a SyncMirror aggregate. To do that, I turn off disk auto assign and I manually assign half of my disks that were in Pool 0 to Pool 1.
Next, I convert Aggregate 1 into a SyncMirror aggregate. When I do that I still have the original disk shelf. The data is on there in Pool 0 and Plex 0. I'm now going to be mirroring the data to a different disk shelf, so it's an exact mirror copy of Plex 0.
Now I have Plex 0 on one shelf and Plex 1 on the other shelf. Notice that I have half of my spare disks in Pool 0 and the other half of my spare disks in Pool 1.
In the example here, I converted an existing aggregate into a SyncMirror aggregate. I can also do this for new aggregates when I create them, as long as I have enough available disks in Pool 0 and Pool 1.
Data mirroring using SyncMirror from NetApp.
Text by Alex Papas, Technical Writer at www.flackbox.com
Alex has been working with Data Center technologies for over 20 years. Currently he is the Network Lead for Costa, one of the largest agricultural companies in Australia. When he’s not knee deep in technology you can find Alex performing with his band 2am