When creating storage accounts in Azure there are numerous options to consider for redundancy, and there are trade-offs to be made between availability and cost.
The below table summarises the different outage scenarios and the level of protection given for each option.
|Loss of node within data centre||Yes||Yes||Yes||Yes||Yes||Yes|
|Loss of data centre||No||Yes||Yes||Yes||Yes||Yes|
|Loss of primary region||No||No||Yes||Yes||Yes||Yes|
|Read access to secondary region||No||No||No||No||Yes||Yes|
Each of the different options are broken down as follows:
- Locally Redundant Storage (LRS) – Data is synchronously replicated three times within a data centre in the primary region.
- Zone Redundant Storage (ZRS) – Data is synchronously replicated across three availability zones in the primary region.
- Geo Redundant Storage (GRS) – Data is synchronously replicated within a single data centre in the primary region (as per LRS). The data is then asynchronously replicated to a data centre in the secondary region and in addition this data replicated three times within the data centre.
- Geo-Zone Redundant Storage (GZRS) – Data is synchronously replicated across three availability zones in the primary region and then asynchronously replicated to the secondary region, data in secondary region is replicated three times within the same datacentre.
- RA-GRS and RA-GZRS provide read-only access to the data in the secondary region without the need to initiate a failover from the primary region. In effect, this provides greater control over the Recovery Time Objective (RTO) as you can read the data from the secondary endpoint as well as the primary endpoint.
As LRS provides the most basic level of protection it is also the most cost effective. Each step up in redundancy represents an increase in cost, so it’s important to understand what the acceptable level of risk is versus cost.