This is part of a series on my preparation for the DP-900 exam. This is the Microsoft Azure Data Fundamentals, part of a number of certification paths. You can read various posts I’ve created as part of this learning experience.
In a previous post, I discussed the makeup of Azure regions and Availability Zones. These are important concepts to understand how storage protection works. This post looks at the various redundancy types: LRS, ZRS, GRS, and GZRS and RA-GZRS.
Note: This isn’t directly on DP-900, but these concepts did help me understand better how the Azure options fit together.
As I discuss these, this infographic might help. Also, MS Docs has an article.
Local Redundant Storage – LRS
As it might imply, this means that your storage bits are copied synchronously to redundant storage in the local area. In this case, this means the data center where your storage is located. Your files (or VHD or whatever) is copied three times into separate locations. This means that your files can survive a single node (computer or disk) failure.
This also means you should forego software RAID for VMs or data files as there are already three copies that will acknowledge the write.
You do not get to pick which data center, just the region where the data center lives. This means your resource could be in any of the data centers in a region, but there are three copies.
Impressively, this gives you 11 9s of durability. However, a fire or major failure of a single data center might cause all your data to disappear.
You can use managed disks with LRS.
Zone Redundant Storage – ZRS
The next level of protection is ZRS, where you get three copies, but this time the copies exist in three parts of an Availability Zone. This is why the AZ must be inside a 2ms latency boundary. This way your files can be copied from data center to data center inside of the AZ.
In this way you achieve 12 9s of durability for your files. More importantly, if a single data center were to go offline, you would still have access to your files (or VHDs).
You can use managed disks with ZRS.
Geo-Redundant Storage – GRS
This is a type of redundancy that is between LRS and GZRS. In GRS, the primary data center implements LRS, with 3 copies of files in the single data center. This gives you protection against node failures.
A second copy, asynchronously, is made in a second region, where LRS is also implemented. This means 6 copies of your data, spread across 2 data centers, one in each of two regions. The image below shows this:
Note, the files in the secondary region are not accessible unless there is a failover. However, if you add RA-GRS, the read access equivalent, you can access files in either zone.
Geo-Zone Redundant Storage – GZRS
GZRS expands the ZRS storage to include a second region. You still get three copies in your primary region spread out across an AZ, but you also get three copies in a secondary region. These are replicated asynchronously as the latency between regions isn’t bounded closely enough for synchronous changes.
The secondary zone uses LRS for the copies. The difference between GRS and GZRS is the replication in the primary region.
GZRS is supposed to give you 16 9s of durability.
Read Access Geo-Zone Redundant Storage – RA-GZRS
With GZRS, there is more protection for your files, but you cannot access the files in the secondary region unless you fail over from the primary region. The customer or Microsoft can initiate this.
In some cases, you might wish that you could spread the read workload between the primary and secondary regions. With RA-GZRS, you can do this. Your clients can access files in the secondary or primary region.
Availability of Azure Storage Options
Not all Azure Storage options are available with the different redundancy options. In the redundancy article, there are two charts, I’ve reproduced below. This first one looks at the different redundancy options and which types of storage fit:
This next one shows that the different account types fitting in the options.