Data redundancy in Azure storage helps to ensure the availability and durability of data services in case of any mishaps.
Multiple factors such as budget, availability requirements, data replication strategy, and geo-restrictions can influence determining the best redundancy options.
However, whatever option you go with; the redundancy settings will be applicable for all storage services created using the storage account. If there’s a need to have separate redundancy for a specific group of data storage services; it’s better to have a separate storage account created for those.
Data redundancies in Microsoft Azure can be broadly classified into two categories:
Primary region redundancies:
There are two redundancy options available in the primary region:
1. Locally redundant storage (LRS)
It’s the least expensive option and it stores three copies of data in a single physical location in the primary region. It’s a high-risk option if available, durability are the key criterion.
LRS provides yearly SLA of at least 11 9s – 99.999999999%. It protects data from server racks and drives failures but data center-level failures can cause permanent data loss.
LRS can be used if the data can be easily retrieved from local copies or re-created and if there’s a restriction of storing data in a specific region only.
2. Zone-redundant storage (ZRS)
It replicates data to three Azure availability zones in the primary region.
ZRS provides yearly SLA of at least 12 9s – 99.9999999999%.
ZRS can be used in high-availability scenarios within the primary region and also with region or country-specific data storage restrictions.
If an availability zone becomes unavailable then Microsoft Azure does DNS repointing to available zones.
Secondary region redundancies
The paired secondary region is auto-determined and controlled (can’t be changed) by Microsoft Azure based on the primary region.
There are two options available to copy data to a secondary region:
1. Geo-redundant storage (GRS)
Post creating three replicas in a single physical location within the primary region; Azure uses LRS to asynchronously create three replicas again in a single physical location within the secondary region.
GRS provides yearly SLA of at least 16 9s – 99.99999999999999%. It protects data from server racks and drives failures but data center-level failures can cause permanent data loss.
LRS can be used if the data can be easily retrieved from local copies or re-created and if there’s a restriction of storing data in a specific region only.
2. Geo-zone-redundant storage (GZRS)
Synchronously replicates data across 3 availability zones within the primary region by using ZRS. After that, data is asynchronously copied to a single physical location three times using LRS within the secondary region.
GZRS provides yearly SLA of at least 16 9s – 99.99999999999999%
3. Read Access – Geo-redundant storage (RA-GRS)
Makes the secondary region available for read-access thus enabling both regions to provide read-access capability through primary and secondary endpoints.
4. Read Access – Geo-zone-redundant storage (RA-GZRS)
Same principles as RA-GRS
References:
https://learn.microsoft.com/en-us/azure/storage/common/storage-redundancy
