Replication and redundancy

In order to guarantee stability and high availability, the customer's data in the Azure data center is replicated constantly. Before I introduce you to the mechanisms that Azure offers, first let's look at some theory:

• Replication refers to the multiple storage of the same data at mostly several different sites and the synchronization of these data sources.
• Redundancy is the additional presence of functionally identical or comparable resources of a technical system, which are not required where normal operation is trouble-free. In other words, redundancy is a direct result of successful replication.

We now come to the mechanisms already mentioned. The Azure platform offers you:

• Locally redundant storage (LRS): LRS means that the data is held three times in one data center from one region. The LRS manages three copies of the customer's data to protect it from hardware failures. LRS does not protect the workloads from the failure of a whole data center.
• Zone-redundant storage (ZRS): ZRS stores three copies of the customer's data as well as the LRS. The difference is that the data is guarded in two to three facilities. These facilities can be located in the same or in a different region. This concept provides more enhanced durability than LRS. The user profits from durability within a region.
• Geo-redundant storage (GRS): An even higher durability can be achieved with GRS. GRS manages six copies of the user's data. The first three copies are replicated in the primary region. Additionally, another three copies are maintained in a secondary region which is located remotely from the primary region. This concept provides an even higher level of durability. This means that Azure failovers to the secondary region if a failure in the primary region should occur.
• Read-access geo-redundant storage (RA-GRS): The replication to a secondary geographic location is provided with read access. The customer holds read access to the data, maintained in the secondary location. Access from the primary and the secondary region is possible. The RA-GRS is the default option for your storage account on creation.

The mechanisms just described are only valid for Azure Storage services. With the Azure SQL as a service offers, modified options come into action:

• Azure SQL Database, Azure SQL Elastic Database or Azure SQL Data Warehouse first use the Azure Backup service to create a full, differential, or transaction log backup
• Then, in a second step, the backup is stored in a storage account and saved with the mechanisms already described

The backups have a so-called retention time. Depending on the selected performance level, the availability ends after 7 days (standard level) or 35 days (premium level).

Azure PostgreSQL as a service and Azure MySQL as a service (both currently in the preview state) do not have the option of automatic backup. Here, you have to create a manual backup first and then you can save the backup files in the Azure Storage.