Resource limits for single databases using the vCore purchasing model
Applies to: Azure SQL Database
This article provides the detailed resource limits for single databases in Azure SQL Database using the vCore purchasing model.
- For elastic pool vCore resource limits, vCore resource limits - elastic pools.
- For limits of the logical server in Azure, see Overview of resource limits on a server.
- For DTU purchasing model resource limits, see DTU resource limits single databases and DTU resource limits elastic pools.
- For more information regarding the different purchasing models, see Purchasing models and service tiers.
Important
Shrink operations should not be considered a regular maintenance operation. Data and log files that grow due to regular, recurring business operations do not require shrink operations.
- Under some circumstances, you might need to shrink a database to reclaim unused space. For more information, see Manage file space in Azure SQL Database.
- For scaling guidance and considerations, see Scale a single database.
Each read-only replica of a database has its own resources, such as vCores, memory, data IOPS, tempdb
, workers, and sessions. Each read-only replica is subject to the resource limits detailed later in this article.
You can set the service tier, compute size (service objective), and storage amount for a single database using:
Note
The Gen5 hardware in the vCore purchasing model has been renamed to standard-series (Gen5).
General Purpose - serverless compute - standard-series (Gen5)
The serverless compute tier is currently available on standard-series (Gen5) hardware only.
General Purpose - serverless compute - standard-series (Gen5) (part 1 of 3)
Compute sizes (service level objectives, or SLOs) for General Purpose serverless standard-series databases follow the naming convention GP_S_Gen5_
followed by the number of max vCores.
The following table covers these SLOs: GP_S_Gen5_1
, GP_S_Gen5_2
, GP_S_Gen5_4
, GP_S_Gen5_6
, and GP_S_Gen5_8
:
Min-max vCores | 0.5 - 1 | 0.5 - 2 | 0.5 - 4 | 0.75 - 6 | 1.0 - 8 |
---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Min-max memory (GB) | 2.02 - 3 | 2.05 - 6 | 2.10 - 12 | 2.25 - 18 | 3.00 - 24 |
Min-max auto-pause delay (minutes) | 15 - 10,080 | 15 - 10,080 | 15 - 10,080 | 15 - 10,080 | 15 - 10,080 |
Columnstore support | Yes 1 | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A |
Max data size (GB) | 512 | 1024 | 1024 | 1024 | 2048 |
Max log size (GB) 2 | 154 | 307 | 307 | 307 | 614 |
Tempdb max data size (GB) | 32 | 64 | 128 | 192 | 256 |
Storage type | Remote SSD | Remote SSD | Remote SSD | Remote SSD | Remote SSD |
Read IO latency (approximate) | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms |
Write IO latency (approximate) | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms |
Max data IOPS 3 | 320 | 640 | 1280 | 1920 | 2560 |
Max log rate (MBps) | 4.5 | 9 | 18 | 27 | 36 |
Max concurrent workers | 75 | 150 | 300 | 450 | 600 |
Max concurrent logins | 75 | 150 | 300 | 450 | 600 |
Max concurrent external connections 4 | 7 | 15 | 30 | 45 | 60 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 1 | 1 | 1 | 1 | 1 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | N/A | N/A | N/A | N/A | N/A |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 Service objectives with smaller max vCore configurations can have insufficient memory for creating and using columnstore indexes. If encountering performance problems with columnstore, increase the max vCore configuration to increase the max memory available.
2 For documented max data size values. Reducing max data size reduces max log size proportionally.
3 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
4 For more information on what counts as an external connection, see External Connections.
General Purpose - serverless compute - standard-series (Gen5) (part 2 of 3)
Compute sizes (service level objectives, or SLOs) for General Purpose serverless standard-series databases follow the naming convention GP_S_Gen5_
followed by the number of max vCores.
The following table covers these SLOs: GP_S_Gen5_10
, GP_S_Gen5_12
, GP_S_Gen5_14
, GP_S_Gen5_16
, and GP_S_Gen5_18
:
Min-max vCores | 1.25-10 | 1.50-12 | 1.75-14 | 2.00-16 | 2.25-18 |
---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Min-max memory (GB) | 3.75-30 | 4.50-36 | 5.25-42 | 6.00-48 | 6.75-54 |
Min-max auto-pause delay (minutes) | 15-10,080 | 15-10,080 | 15-10,080 | 15-10,080 | 15-10,080 |
Columnstore support | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A |
Max data size (GB) | 2048 | 3072 | 3072 | 3072 | 3072 |
Max log size (GB) 1 | 614 | 922 | 922 | 922 | 922 |
Tempdb max data size (GB) | 320 | 384 | 448 | 512 | 576 |
Storage type | Remote SSD | Remote SSD | Remote SSD | Remote SSD | Remote SSD |
Read IO latency (approximate) | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms |
Write IO latency (approximate) | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms |
Max data IOPS 2 | 3200 | 3840 | 4480 | 5120 | 5760 |
Max log rate (MBps) | 45 | 50 | 50 | 50 | 50 |
Max concurrent workers | 750 | 900 | 1050 | 1200 | 1350 |
Max concurrent logins | 750 | 900 | 1050 | 1200 | 1350 |
Max concurrent external connections 3 | 75 | 90 | 105 | 120 | 135 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 1 | 1 | 1 | 1 | 1 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | N/A | N/A | N/A | N/A | N/A |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
General Purpose - serverless compute - standard-series (Gen5) (part 3 of 3)
Compute sizes (service level objectives, or SLOs) for General Purpose serverless standard-series databases follow the naming convention GP_S_Gen5_
followed by the number of max vCores.
The following table covers these SLOs: GP_S_Gen5_20
, GP_S_Gen5_24
, GP_S_Gen5_32
, and GP_S_Gen5_40
:
Min-max vCores | 2.5-20 | 3-24 | 4-32 | 5-40 |
---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 |
Min-max memory (GB) | 7.5-60 | 9-72 | 12-96 | 15-120 |
Min-max auto-pause delay (minutes) | 15-10,080 | 15-10,080 | 15-10,080 | 15-10,080 |
Columnstore support | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A |
Max data size (GB) | 3072 | 4096 | 4096 | 4096 |
Max log size (GB) 1 | 922 | 1024 | 1024 | 1024 |
Tempdb max data size (GB) | 640 | 768 | 1024 | 1280 |
Storage type | Remote SSD | Remote SSD | Remote SSD | Remote SSD |
Read IO latency (approximate) | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms |
Write IO latency (approximate) | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms |
Max data IOPS 2 | 6,400 | 7,680 | 10,240 | 12,800 |
Max log rate (MBps) | 50 | 50 | 50 | 50 |
Max concurrent workers | 1500 | 1800 | 2400 | 3000 |
Max concurrent logins | 1500 | 1800 | 2400 | 3000 |
Max concurrent external connections 3 | 150 | 150 | 150 | 150 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 1 | 1 | 1 | 1 |
Multi-AZ | Yes | Yes | Yes | Yes |
Read Scale-out | N/A | N/A | N/A | N/A |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
Hyperscale - serverless compute - standard-series (Gen5)
The serverless compute tier is currently available on standard-series (Gen5) hardware only.
Hyperscale - serverless compute - standard-series (Gen5) (part 1 of 2)
Compute sizes (service level objectives, or SLOs) for Hyperscale serverless standard-series databases follow the naming convention HS_S_Gen5_
followed by the number of max vCores.
The following table covers these SLOs: HS_S_Gen5_2
, HS_S_Gen5_4
, HS_S_Gen5_6
, HS_S_Gen5_8
, HS_S_Gen5_10
, HS_S_Gen5_12
, and HS_S_Gen5_14
:
Min-max vCores | 0.5-2 | 0.5-4 | 0.75-6 | 1-8 | 1.25-10 | 1.5-12 | 1.75-14 |
---|---|---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Min-max memory (GB) | 2.05-6 | 2.10-12 | 2.25-18 | 3.00-24 | 3.75-30 | 4.50-36 | 5.25-42 |
Columnstore support | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A | N/A | N/A |
Max data size (TB) | 128 | 128 | 128 | 128 | 128 | 128 | 128 |
Max log size (TB) | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited |
Tempdb max data size (GB) | 64 | 128 | 192 | 256 | 320 | 384 | 448 |
Storage type | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 |
Max local SSD IOPS 2 | 8000 | 16000 | 24000 | 32000 | 40000 | 48000 | 56000 |
Max log rate (MB/s) | 100 | 100 | 100 | 100 | 100 | 100 | 100 |
IO latency (approximate) | Variable 3 | Variable 3 | Variable 3 | Variable 3 | Variable 3 | Variable 3 | Variable 3 |
Max concurrent workers | 150 | 300 | 450 | 600 | 750 | 900 | 1050 |
Max concurrent logins | 150 | 300 | 450 | 600 | 750 | 900 | 1050 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Secondary replicas | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 |
Multi-AZ | N/A | N/A | N/A | N/A | N/A | N/A | N/A |
Read Scale-out | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Backup storage retention | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days |
1 Hyperscale is a multi-tiered architecture with separate compute and storage components. Review Hyperscale service tier for more information.
2 Besides local SSD IO, workloads use remote page server IO. Actual IOPS are workload-dependent. For details, see Data IO Governance, and Data IO in resource utilization statistics.
3 Latency is 1-2 ms for data on local compute replica SSD, which caches most used data pages. Higher latency for data retrieved from page servers.
Hyperscale - serverless compute - standard-series (Gen5) (part 2 of 2)
Compute sizes (service level objectives, or SLOs) for Hyperscale serverless standard-series databases follow the naming convention HS_S_Gen5_
followed by the number of max vCores.
The following table covers these SLOs: HS_S_Gen5_16
, HS_S_Gen5_18
, HS_S_Gen5_20
, HS_S_Gen5_24
, HS_S_Gen5_32
, HS_S_Gen5_40
, and HS_S_Gen5_80
:
Min-max vCores | 2-16 | 2.25-18 | 2.25-20 | 3-24 | 4-32 | 5-40 |
---|---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Min-max memory (GB) | 6.00-48 | 6.75-54 | 7.5-60 | 9-72 | 12-96 | 15-120 |
Columnstore support | Yes | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A | N/A |
Max data size (TB) | 128 | 128 | 128 | 128 | 128 | 128 |
Max log size (TB) | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited |
Tempdb max data size (GB) | 512 | 576 | 640 | 768 | 1024 | 1280 |
Storage type | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 | Multi-tiered 1 |
Max local SSD IOPS 2 | 64000 | 72000 | 80000 | 96000 | 128000 | 160000 |
Max log rate (MB/s) | 100 | 100 | 100 | 100 | 100 | 100 |
IO latency (approximate) | Variable 3 | Variable 3 | Variable 3 | Variable 3 | Variable 3 | Variable 3 |
Max concurrent workers | 1200 | 1350 | 1500 | 1800 | 2400 | 3000 |
Max concurrent logins | 1200 | 1350 | 1500 | 1800 | 2400 | 3000 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Secondary replicas | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 |
Multi-AZ | N/A | N/A | N/A | N/A | N/A | N/A |
Read Scale-out | Yes | Yes | Yes | Yes | Yes | Yes |
Backup storage retention | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days |
1 Hyperscale is a multi-tiered architecture with separate compute and storage components. Review Hyperscale service tier for more information.
2 Besides local SSD IO, workloads use remote page server IO. Actual IOPS are workload-dependent. For details, see Data IO Governance, and Data IO in resource utilization statistics.
3 Latency is 1-2 ms for data on local compute replica SSD, which caches most used data pages. Higher latency for data retrieved from page servers.
Hyperscale - provisioned compute - standard-series (Gen5)
Hyperscale standard-series (Gen5) (part 1 of 2)
Compute sizes (service level objectives, or SLOs) for Hyperscale standard-series databases follow the naming convention HS_Gen5_
followed by the number of vCores.
The following table covers these SLOs: HS_Gen5_2
, HS_Gen5_4
, HS_Gen5_6
, HS_Gen5_8
, HS_Gen5_10
, HS_Gen5_12
, and HS_Gen5_14
:
vCores | 2 | 4 | 6 | 8 | 10 | 12 | 14 |
---|---|---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 10.4 | 20.8 | 31.1 | 41.5 | 51.9 | 62.3 | 72.7 |
Columnstore support | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A | N/A | N/A |
Max data size (TB) | 128 | 128 | 128 | 128 | 128 | 128 | 128 |
Max log size (TB) | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited |
Tempdb max data size (GB) |
64 | 128 | 192 | 256 | 320 | 384 | 448 |
Max local SSD IOPS 1 | 8000 | 16,000 | 24,000 | 32,000 | 40,000 | 48,000 | 56,000 |
Max log rate (MB/s) | 100 | 100 | 100 | 100 | 100 | 100 | 100 |
Local read IO latency2 | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Remote read IO latency2 | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms |
Write IO latency2 | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms |
Storage type | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 |
Max concurrent workers | 200 | 400 | 600 | 800 | 1000 | 1200 | 1400 |
Max concurrent logins | 200 | 400 | 600 | 800 | 1000 | 1200 | 1400 |
Max concurrent external connections 4 | 20 | 40 | 60 | 80 | 100 | 120 | 140 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Secondary replicas | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Backup storage retention | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days |
1 Besides local SSD IO, workloads use remote page server IO. Actual IOPS are workload-dependent. For details, see Data IO Governance, and Data IO in resource utilization statistics.
2 Latency numbers are approximate and representative for typical workloads at steady state, but aren't guaranteed.
3 Hyperscale is a multi-tiered architecture with separate compute and storage components. For more information, see Hyperscale service tier architecture.
4 For more information on what counts as an external connection, see External Connections.
Hyperscale standard-series (Gen5) (part 2 of 2)
Compute sizes (service level objectives, or SLOs) for Hyperscale standard-series databases follow the naming convention HS_Gen5_
followed by the number of vCores.
The following table covers these SLOs: HS_Gen5_16
, HS_Gen5_18
, HS_Gen5_20
, HS_Gen5_24
, HS_Gen5_32
, HS_Gen5_40
, and HS_Gen5_80
:
vCores | 16 | 18 | 20 | 24 | 32 | 40 | 80 |
---|---|---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 83 | 93.4 | 103.8 | 124.6 | 166.1 | 207.6 | 415.2 |
Columnstore support | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A | N/A | N/A |
Max data size (TB) | 128 | 128 | 128 | 128 | 128 | 128 | 128 |
Max log size (TB) | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited | Unlimited |
Tempdb max data size (GB) | 512 | 576 | 640 | 768 | 1024 | 1280 | 2560 |
Max local SSD IOPS 1 | 64,000 | 72,000 | 80,000 | 96,000 | 128,000 | 160,000 | 204,800 |
Max log rate (MB/s) | 100 | 100 | 100 | 100 | 100 | 100 | 100 |
Local read IO latency2 | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Remote read IO latency2 | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms |
Write IO latency2 | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms | 1-4 ms |
Storage type | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 | Multi-tiered3 |
Max concurrent workers | 1600 | 1800 | 2000 | 2400 | 3200 | 4000 | 8000 |
Max concurrent logins | 1600 | 1800 | 2000 | 2400 | 3200 | 4000 | 8000 |
Max concurrent external connections 4 | 150 | 150 | 150 | 150 | 150 | 150 | 150 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Secondary replicas | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 | 0-4 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Backup storage retention | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days | 7 days |
1 Besides local SSD IO, workloads use remote page server IO. Actual IOPS are workload-dependent. For details, see Data IO Governance, and Data IO in resource utilization statistics.
2 Latency numbers are representative for typical workloads at steady state, but aren't guaranteed.
3 Hyperscale is a multi-tiered architecture with separate compute and storage components. For more information, see Hyperscale service tier architecture.
4 For more information on what counts as an external connection, see External Connections.
General Purpose - provisioned compute - standard-series (Gen5)
General Purpose standard-series (Gen5) (part 1 of 3)
Compute sizes (service level objectives, or SLOs) for General Purpose standard-series databases follow the naming convention GP_Gen5_
followed by the number of vCores.
The following table covers these SLOs: GP_Gen5_2
, GP_Gen5_4
, GP_Gen5_6
, GP_Gen5_8
, and GP_Gen5_10
:
vCores | 2 | 4 | 6 | 8 | 10 |
---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 10.4 | 20.8 | 31.1 | 41.5 | 51.9 |
Columnstore support | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A |
Max data size (GB) | 1024 | 1024 | 1536 | 2048 | 2048 |
Max log size (GB) 1 | 307 | 307 | 461 | 461 | 461 |
Tempdb max data size (GB) | 64 | 128 | 192 | 256 | 320 |
Storage type | Remote SSD | Remote SSD | Remote SSD | Remote SSD | Remote SSD |
Read IO latency (approximate) | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms |
Write IO latency (approximate) | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms |
Max data IOPS 2 | 640 | 1280 | 1920 | 2560 | 3200 |
Max log rate (MBps) | 9 | 18 | 27 | 36 | 45 |
Max concurrent workers | 200 | 400 | 600 | 800 | 1000 |
Max concurrent logins | 200 | 400 | 600 | 800 | 1000 |
Max concurrent external connections 3 | 20 | 40 | 60 | 80 | 100 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 1 | 1 | 1 | 1 | 1 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | N/A | N/A | N/A | N/A | N/A |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
General Purpose standard-series (Gen5) (part 2 of 3)
Compute sizes (service level objectives, or SLOs) for General Purpose standard-series databases follow the naming convention GP_Gen5_
followed by the number of vCores.
The following table covers these SLOs: GP_Gen5_12
, GP_Gen5_14
, GP_Gen5_16
, GP_Gen5_18
, and GP_Gen5_20
:
vCores | 12 | 14 | 16 | 18 | 20 |
---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 62.3 | 72.7 | 83 | 93.4 | 103.8 |
Columnstore support | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A | N/A |
Max data size (GB) | 3072 | 3072 | 3072 | 3072 | 3072 |
Max log size (GB) 1 | 922 | 922 | 922 | 922 | 922 |
Tempdb max data size (GB) | 384 | 384 | 512 | 576 | 640 |
Storage type | Remote SSD | Remote SSD | Remote SSD | Remote SSD | Remote SSD |
Read IO latency (approximate) | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms |
Write IO latency (approximate) | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms |
Max data IOPS 2 | 3,840 | 4,480 | 5,120 | 5,760 | 6,400 |
Max log rate (MBps) | 50 | 50 | 50 | 50 | 50 |
Max concurrent workers | 1,200 | 1,400 | 1,600 | 1,800 | 2,000 |
Max concurrent logins | 1,200 | 1,400 | 1,600 | 1,800 | 2,000 |
Max concurrent external connections 3 | 120 | 140 | 150 | 150 | 150 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 1 | 1 | 1 | 1 | 1 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | N/A | N/A | N/A | N/A | N/A |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
General Purpose standard-series (Gen5) (part 3 of 3)
Compute sizes (service level objectives, or SLOs) for General Purpose standard-series databases follow the naming convention GP_Gen5_
followed by the number of vCores.
The following table covers these SLOs: GP_Gen5_24
, GP_Gen5_32
, GP_Gen5_40
, and GP_Gen5_80
:
vCores | 24 | 32 | 40 | 80 |
---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 124.6 | 166.1 | 207.6 | 415.2 |
Columnstore support | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | N/A | N/A | N/A | N/A |
Max data size (GB) | 4096 | 4096 | 4096 | 4096 |
Max log size (GB) 1 | 1024 | 1024 | 1024 | 1024 |
Tempdb max data size (GB) | 768 | 1024 | 1280 | 2560 |
Storage type | Remote SSD | Remote SSD | Remote SSD | Remote SSD |
Read IO latency (approximate) | 5-10 ms | 5-10 ms | 5-10 ms | 5-10 ms |
Write IO latency (approximate) | 5-7 ms | 5-7 ms | 5-7 ms | 5-7 ms |
Max data IOPS 2 | 7,680 | 10,240 | 12,800 | 12,800 |
Max log rate (MBps) | 50 | 50 | 50 | 50 |
Max concurrent workers | 2,400 | 3,200 | 4,000 | 8,000 |
Max concurrent logins | 2,400 | 3,200 | 4,000 | 8,000 |
Max concurrent external connections 3 | 150 | 150 | 150 | 150 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 1 | 1 | 1 | 1 |
Multi-AZ | Yes | Yes | Yes | Yes |
Read Scale-out | N/A | N/A | N/A | N/A |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
Business Critical - provisioned compute - standard-series (Gen5)
Business Critical standard-series (Gen5) (part 1 of 3)
Compute sizes (service level objectives, or SLOs) in the Business Critical standard-series databases follow the naming convention BC_Gen5_
followed by the number of vCores.
The following table covers these SLOs: BC_Gen5_2
, BC_Gen5_4
, BC_Gen5_6
, BC_Gen5_8
, and BC_Gen5_10
:
vCores | 2 | 4 | 6 | 8 | 10 |
---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 10.4 | 20.8 | 31.1 | 41.5 | 51.9 |
Columnstore support | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | 1.57 | 3.14 | 4.71 | 6.28 | 8.65 |
Max data size (GB) | 1024 | 1024 | 1536 | 2048 | 2048 |
Max log size (GB) 1 | 307 | 307 | 461 | 461 | 461 |
Tempdb max data size (GB) | 64 | 128 | 192 | 256 | 320 |
Max local storage size (GB) | 4829 | 4829 | 4829 | 4829 | 4829 |
Storage type | Local SSD | Local SSD | Local SSD | Local SSD | Local SSD |
Read IO latency (approximate) | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Write IO latency (approximate) | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Max data IOPS 2 | 8000 | 16,000 | 24,000 | 32,000 | 40,000 |
Max log rate (MBps) | 24 | 48 | 72 | 96 | 96 |
Max concurrent workers | 200 | 400 | 600 | 800 | 1000 |
Max concurrent logins | 200 | 400 | 600 | 800 | 1000 |
Max concurrent external connections 3 | 20 | 40 | 60 | 80 | 100 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 4 | 4 | 4 | 4 | 4 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | Yes | Yes | Yes | Yes | Yes |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
Business Critical standard-series (Gen5) (part 2 of 3)
Compute sizes (service level objectives, or SLOs) in the Business Critical standard-series databases follow the naming convention BC_Gen5_
followed by the number of vCores.
The following table covers these SLOs: BC_Gen5_12
, BC_Gen5_14
, BC_Gen5_16
, BC_Gen5_18
, and BC_Gen5_20
:
vCores | 12 | 14 | 16 | 18 | 20 |
---|---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 62.3 | 72.7 | 83 | 93.4 | 103.8 |
Columnstore support | Yes | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | 11.02 | 13.39 | 15.77 | 18.14 | 20.51 |
Max data size (GB) | 3072 | 3072 | 3072 | 3072 | 3072 |
Max log size (GB) 1 | 922 | 922 | 922 | 922 | 922 |
Tempdb max data size (GB) | 384 | 448 | 512 | 576 | 640 |
Max local storage size (GB) | 4829 | 4829 | 4829 | 4829 | 4829 |
Storage type | Local SSD | Local SSD | Local SSD | Local SSD | Local SSD |
Read IO latency (approximate) | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Write IO latency (approximate) | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Max data IOPS 2 | 48,000 | 56,000 | 64,000 | 72,000 | 80,000 |
Max log rate (MBps) | 96 | 96 | 96 | 96 | 96 |
Max concurrent workers | 1200 | 1400 | 1600 | 1800 | 2000 |
Max concurrent logins | 1200 | 1400 | 1600 | 1800 | 2000 |
Max concurrent external connections 3 | 120 | 140 | 150 | 150 | 150 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 4 | 4 | 4 | 4 | 4 |
Multi-AZ | Yes | Yes | Yes | Yes | Yes |
Read Scale-out | Yes | Yes | Yes | Yes | Yes |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
Business Critical standard-series (Gen5) (part 3 of 3)
Compute sizes (service level objectives, or SLOs) in the Business Critical standard-series databases follow the naming convention BC_Gen5_
followed by the number of vCores.
The following table covers these SLOs: BC_Gen5_24
, BC_Gen5_32
, BC_Gen5_40
, and BC_Gen5_80
:
vCores | 24 | 32 | 40 | 80 |
---|---|---|---|---|
Hardware | Gen5 | Gen5 | Gen5 | Gen5 |
Memory (GB) | 124.6 | 166.1 | 207.6 | 415.2 |
Columnstore support | Yes | Yes | Yes | Yes |
In-memory OLTP storage (GB) | 25.25 | 37.94 | 52.23 | 131.64 |
Max data size (GB) | 4096 | 4096 | 4096 | 4096 |
Max log size (GB) 1 | 1024 | 1024 | 1024 | 1024 |
Tempdb max data size (GB) | 768 | 1024 | 1280 | 2560 |
Max local storage size (GB) | 4829 | 4829 | 4829 | 4829 |
Storage type | Local SSD | Local SSD | Local SSD | Local SSD |
Read IO latency (approximate) | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Write IO latency (approximate) | 1-2 ms | 1-2 ms | 1-2 ms | 1-2 ms |
Max data IOPS 2 | 96,000 | 128,000 | 160,000 | 204,800 |
Max log rate (MBps) | 96 | 96 | 96 | 96 |
Max concurrent workers | 2400 | 3200 | 4000 | 8000 |
Max concurrent logins | 2400 | 3200 | 4000 | 8000 |
Max concurrent external connections 3 | 150 | 150 | 150 | 150 |
Max concurrent sessions | 30,000 | 30,000 | 30,000 | 30,000 |
Number of replicas | 4 | 4 | 4 | 4 |
Multi-AZ | Yes | Yes | Yes | Yes |
Read Scale-out | Yes | Yes | Yes | Yes |
Included backup storage | 1X DB size | 1X DB size | 1X DB size | 1X DB size |
1 For documented max data size values. Reducing max data size reduces max log size proportionally.
2 The maximum value for IO sizes ranging between 8 KB and 64 KB. Actual IOPS are workload-dependent. For details, see Data IO Governance.
3 For more information on what counts as an external connection, see External Connections.
Previously available hardware
This section includes details on previously available hardware.
Gen4 hardware has been retired and isn't available for provisioning, upscaling, or downscaling. Migrate your database to a supported hardware generation for a wider range of vCore and storage scalability, accelerated networking, best IO performance, and minimal latency. For more information, see Azure SQL Database Gen 4 hardware approaching end of life.
Next steps
- For DTU resource limits for a single database, see resource limits for single databases using the DTU purchasing model
- For vCore resource limits for elastic pools, see resource limits for elastic pools using the vCore purchasing model
- For DTU resource limits for elastic pools, see resource limits for elastic pools using the DTU purchasing model
- For resource limits for SQL Managed Instance, see SQL Managed Instance resource limits.
- For information about general Azure limits, see Azure subscription and service limits, quotas, and constraints.
- For information about resource limits at the server and subscription levels, see overview of resource limits on a server.
- ALTER DATABASE - Azure SQL Database
- CREATE DATABASE - Azure SQL Database