Azure SQL Database Hyperscale
A next-generation cloud database architecture introduced by Microsoft in 2018 that fully decouples the compute and storage layers. Storage capacity can scale beyond 100 TB, read replicas can be added rapidly, and pricing is comparable to mainstream open-source databases.
Timeline (last 90 days)
The Hyperscale compute layer can scale up to 192 vCores, and scaling compute does not require moving the database.
Hyperscale supports up to 30 named replicas, each of which can be configured independently with compute resources to isolate read-intensive workloads.
Hyperscale offers an availability SLA of up to 99.995% when configured with zone redundancy.
The Hyperscale log service can provide transaction log throughput of up to 150 MB/s, independent of the compute size of the primary replica.
A named replica is a read-only compute replica in Hyperscale used for read scale-out, with its own connection endpoint, that does not participate in automatic failover and shares the primary replica's page servers and storage layer.
A page server is a core storage component in the Hyperscale architecture that manages a portion of the database's data pages and maintains its own SSD cache, consumes transaction log records and persists updates to Azure Storage.
Resilient Buffer Pool Extension is a local SSD caching mechanism on Hyperscale compute replicas used to extend memory buffer pool capacity and reduce remote requests to page servers.
Hyperscale can be configured with 0 to 4 high-availability replicas as hot standby compute for automatic failover, and adding a high-availability replica does not require creating another full copy of the database.
Hyperscale backup and restore are performed using file snapshots at the storage layer, and same-region restore can complete within minutes, with restore time no longer growing linearly with database size.
The Hyperscale Premium series configuration can reach up to 192 vCores, 843 GB of memory and 816,000 local SSD IOPS.
1 more timeline events
All Facts (13)
Hyperscale's Continuous Priming mechanism tracks active pages accessed by the primary replica, which are fed back to high-availability replicas by the page servers to shorten performance recovery time after failover.
50%UnverifiedHyperscale backup and restore are performed using file snapshots at the storage layer, and same-region restore can complete within minutes, with restore time no longer growing linearly with database size.
50%UnverifiedThe Hyperscale Premium series configuration can reach up to 192 vCores, 843 GB of memory and 816,000 local SSD IOPS.
50%UnverifiedThe Hyperscale compute layer can scale up to 192 vCores, and scaling compute does not require moving the database.
50%UnverifiedHyperscale supports up to 30 named replicas, each of which can be configured independently with compute resources to isolate read-intensive workloads.
50%UnverifiedHyperscale offers an availability SLA of up to 99.995% when configured with zone redundancy.
50%UnverifiedThe Hyperscale log service can provide transaction log throughput of up to 150 MB/s, independent of the compute size of the primary replica.
50%UnverifiedA named replica is a read-only compute replica in Hyperscale used for read scale-out, with its own connection endpoint, that does not participate in automatic failover and shares the primary replica's page servers and storage layer.
50%UnverifiedA page server is a core storage component in the Hyperscale architecture that manages a portion of the database's data pages and maintains its own SSD cache, consumes transaction log records and persists updates to Azure Storage.
50%UnverifiedResilient Buffer Pool Extension is a local SSD caching mechanism on Hyperscale compute replicas used to extend memory buffer pool capacity and reduce remote requests to page servers.
50%UnverifiedHyperscale can be configured with 0 to 4 high-availability replicas as hot standby compute for automatic failover, and adding a high-availability replica does not require creating another full copy of the database.
50%UnverifiedHyperscale 数据库存储容量理论上可扩展至100TB以上
50%UnverifiedAzure SQL Database Hyperscale 是微软于2018年推出的新一代云数据库架构
50%