So that hardware failures, datacentre outages or a regional disaster cannot destroy data, Azure Storage keeps multiple copies: only in the primary region with LRS and ZRS, or in a secondary region as well with GRS and GZRS.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Redundancy in context, with comparison tables and the common traps.
Terms in this definition
- Datacenter
A building packed with racked servers that has its own electricity supply, cooling and network links. Microsoft gathers these buildings into Azure regions, and as a customer you pick the region, never which building your resources land in.
- Nonzonal
Describes a resource with no availability zone configuration. Azure can put it in any zone, so it may go offline if that zone fails.
- Azure Storage
You reach this Azure data storage platform through a storage account; it covers blobs (Data Lake Storage included), file shares, queues and tables.
- region
A provider-defined grouping in VCF Automation of Supervisors that all share one NSX Local Manager; tenants consume its compute, storage and memory via quotas set per region.
- LRS
Locally redundant storage, the lowest-cost redundancy option, keeps three copies of the data inside a single datacentre in the primary region.
- ZRS
Zone-redundant storage: data is written three times, each copy in a different availability zone of the same region, and nothing is replicated to a secondary region.
- Secondary region
Holds disaster-recovery copies of data or workloads, typically hundreds of miles away from the primary location. Geo-redundant storage always uses the primary's pair for this, with no option to change it.
- GRS
Geo-redundant storage keeps six copies in total: three locally redundant in the primary region and three in the paired region. The secondary copy can be read only once a failover has happened.
Related terms
- Allowed storage account SKUs
A built-in Azure Policy that limits the SKUs people may pick for a new storage account. It keeps redundancy and spending in check when a workload has no need for the pricier options.
- Application Gateway v2
Present-day Application Gateway SKU, Standard_v2 or WAF_v2, offering zone redundancy, autoscaling, a static VIP, Key Vault integration and header rewrite. It requires its own subnet, ideally a /24.
- Azure Load Balancer
Layer-4 load balancer operating within a region, with zone redundancy on the Standard SKU; it has no WAF, doesn't terminate TLS and can't route by URL.
- BlockBlobStorage
Premium, SSD-backed account kind dedicated to block and append blobs, giving the lowest latency; redundancy is limited to LRS or ZRS and access tiers aren't available.
- FileStorage
Storage account kind on premium SSD dedicated to Azure Files, supporting only ZRS or LRS redundancy.
- Folder target
Where a DFS namespace folder actually leads: a share's UNC path. Several of them on one folder provide redundancy.
- Managed disks
Block storage disks for VMs whose underlying storage accounts Azure looks after. The only redundancy choices are LRS and ZRS.
- Redundancy conversion
Changing a storage account's redundancy, LRS to ZRS for example, without downtime, started by the customer or by support in regions that allow it. Accounts on general-purpose v1 need upgrading beforehand.