
Introduction
Your Azure SQL Database can hit a storage ceiling while CPU and memory sit mostly idle. The limit depends on the purchasing model, the service tier, the hardware, and sometimes the SKU inside the tier.
This guide covers Azure SQL Database only: DTU, provisioned vCore, serverless, elastic pools, and Hyperscale. Azure SQL Managed Instance and SQL Server on Azure VMs are out of scope.
You'll get the current limits, a way to confirm your active limit, steps to find the real bottleneck, and the least disruptive fix. Verify every figure against Microsoft's documentation before you change tiers, because limits and prices change.
Key Takeaways
- Max size depends on tier, purchasing model, hardware, and SKU.
- Configured maximum, used space, log space, and backup storage are separate quantities.
- Hyperscale reaches 128 TB; General Purpose and Business Critical top out at 4 TB.
- Reclaim data or raise MAXSIZE first, then change tier, then redesign.
- Check Microsoft's current limits and pricing before any change.
Azure SQL Database Max Size by Tier
Figures below come from Microsoft's documentation and can change over time. Older articles often quote outdated numbers, so confirm on the linked Microsoft pages before deploying.
| Model / tier | Single database maximum | Elastic pool maximum | Storage billing |
|---|---|---|---|
| DTU Basic | 2 GB | 156 GB (pool) | Included |
| DTU Standard | 250 GB (S0-S2), 1,024 GB (S3-S12) | Up to 4,096 GB, depending on eDTUs | Allowance included, extra billed |
| DTU Premium | 1,024 GB (P1-P6), 4,096 GB (P11/P15) | Up to 4,096 GB | Allowance included, extra billed |
| vCore General Purpose | Up to 4 TB, SKU-dependent | Up to 4,096 GB | Configured maximum |
| vCore Business Critical | Up to 4 TB, SKU-dependent | Up to 4,096 GB | Configured maximum |
| GP serverless | Up to 4,096 GB | Single databases only | Same as provisioned GP |
| Hyperscale | 128 TB, automatic growth | 100 TB | Allocated data, 10 GB minimum |

The DTU model
Storage is bundled into each DTU tier, so moving up a level changes more than compute. Basic caps at 2 GB. Standard S0-S2 cap at 250 GB, while S3-S12 reach 1,024 GB. Premium P11 and P15 reach 4,096 GB, and some regions exclude Premium databases above 1 TB.
Extra provisioned storage beyond the included allowance is billed separately where eligible. Pool ceilings and per-database ceilings differ, so check both.
The vCore model
General Purpose uses remote premium storage and Business Critical uses local SSD. Both let you set a maximum from 1 GB up to the SKU ceiling in 1 GB steps. On Gen5 General Purpose, 2-4 vCores allow 1,024 GB, 8-10 vCores allow 2,048 GB, and 24 or more vCores allow 4,096 GB.
Serverless follows the same storage rules and billing as provisioned General Purpose. Storage stays billable while the database is paused. The listed limit applies to data capacity, and log space is tracked separately.
Hyperscale for very large databases
Hyperscale separates compute, page servers, a log service, and Azure Storage, so storage scales independently of compute. A single database grows automatically up to 128 TB. You can't set a maximum size.
Considerations:
- Up to 4 HA replicas and 30 named replicas, plus geo-replicas
- No memory-optimized tables and no DBCC CHECKDB
- Billing is on allocated data, with replica compute extra
- Reverse migration is limited, covered below
How Azure SQL Database Storage Limits Work
Several storage figures are easy to mix up. Keep them apart:
- Used: space occupied by data
- Allocated: data-file space, including unused space; deletes don't shrink it automatically
- Configured maximum: the data ceiling you set; allocated data can't exceed it
- Log: separate transaction-log space
- PITR / LTR: separate backup storage
Hitting the data limit differs from hitting log-rate, I/O, CPU, or worker limits. Log-rate governance shows up as LOG_RATE_GOVERNOR waits. On Premium and Business Critical, data, log, and tempdb share local storage, so you can run out below the data-size ceiling.
MAXSIZE is a ceiling, not a performance guarantee. Index rebuilds, staging tables, and retained history all eat headroom. More storage won't fix a log-rate or I/O bottleneck.

How to Diagnose an Azure SQL Database Near Its Maximum Size
Start in the portal
Record these values before you change anything:
- Service tier and purchasing model
- Configured maximum size
- Current usage
- Growth trend
Run T-SQL checks
Per-file allocated versus used space:
SELECT name, size/128.0 AS allocated_MB,
FILEPROPERTY(name,'SpaceUsed')/128.0 AS used_MB
FROM sys.database_files WHERE type_desc='ROWS';
Other documented sources:
sys.dm_db_file_space_usagefor user objects, internal objects, and version-store pagessys.dm_db_partition_statsjoined tosys.objects, usingSUM(reserved_page_count), to find the largest tables and indexessys.dm_db_log_space_usagefor total and used log bytes and percentagesys.dm_db_resource_statsfor CPU, data I/O, log I/O, and workers in 15-second samples, kept for roughly an hour
Use Azure Monitor
Compare metrics side by side rather than reacting to one:
storageandstorage_percentfor data usedallocated_data_storagefor allocationcpu_percent,physical_data_read_percent,log_write_percent, andworkers_percentfor other limits
storage_percent measures against your configured limit, not necessarily the tier ceiling.
Find the fastest growth
Common drivers of rapid growth:
- Large indexes
- Retained audit or event data
- Staging tables
- Unbounded history tables
- Uncompressed large-value columns
- Failed cleanup jobs
Plan capacity
Forecast growth from observed history, add a tested safety margin, and allow for maintenance operations. Compare the result with the tier's documented maximum and your budget.

Capture these so you can judge the fix objectively:
- Before-and-after used space
- Configured maximum
- Workload metrics
- Monthly cost
Fixes for Azure SQL Database Storage Limit Problems
Create immediate headroom
Delete or archive obsolete data, drop abandoned staging tables, and review duplicate or oversized indexes. Deleting doesn't shrink files on its own. DBCC SHRINKFILE can reclaim allocation, but Microsoft doesn't treat it as routine maintenance. It's resource-intensive and can fragment indexes, and later rebuilds can grow allocation again.
Manage the data lifecycle
- Partition history-heavy tables and switch old partitions to archive tables or truncate them
- Export cold data to BACPAC in Blob storage (200 GB maximum for Blob BACPAC; it isn't a backup replacement)
- Move analytical history to a warehouse
- Keep only what applications, regulators, and recovery need
Optimize schema and workload
ROW or PAGE compression cuts rowstore pages and I/O at some CPU cost, though off-row large-value data stays uncompressed. Columnstore is already compressed, and COLUMNSTORE_ARCHIVE saves more with extra CPU. On DTU, columnstore needs Standard S3 or higher, or Premium. Batch ETL jobs to avoid duplicate copies and excess log generation.
Raise the configured maximum
If your current SKU supports more, increase MAXSIZE. Compute upgrades don't raise it automatically. Confirm the tier and region ceiling, check billing impact, and monitor the operation. In pools, a reduced MAXSIZE must still cover total allocated space.
Scale up or change tier
When you've reached the tier ceiling, scaling is the answer. Online preparation is followed by a connection drop that generally lasts under 30 seconds, though long transactions can extend it. Compare capacity, latency, availability, backup behavior, migration effort, and total cost, not storage alone.

Redesign when one database is too big
Consider archiving, sharding, separate read and analytical workloads, multiple databases, external object storage, or a warehouse. Document the consistency, query, backup, and operational trade-offs.
Model costs first
Use the Azure pricing calculator with your real usage. Include compute, data storage, backup storage, retention, redundancy, replicas, and license or reservation discounts. General Purpose and Business Critical bill on configured maximum, while Hyperscale bills on allocated data.
Choosing the Right Azure SQL Database Tier or Architecture
Weigh these factors:
- Database size today and projected growth
- Workload type and write volume
- Latency and availability needs
- Read scale and recovery objectives
- Migration constraints and budget
Microsoft positions the service tiers this way:
- General Purpose for budget-oriented deployments
- Business Critical for high-transaction OLTP with low I/O latency
- Hyperscale as the recommended default for new and modernizing OLTP and HTAP workloads
Hyperscale fits when growth or operational needs outgrow conventional tiers. It's unnecessary for a small database that mainly needs better query or I/O performance.
Plan the exit before you enter. Reverse migration works only for converted databases, within 45 days, to a General Purpose single database with at least 2 vCores. Databases created as Hyperscale are ineligible.
Before any production move:
- Copy the database, then convert the copy
- Run a controlled workload test
- Document a rollback plan

Convert during low-write periods and use retry logic.
Lucidity doesn't raise Azure SQL Database's managed service limit. Teams that also run Azure VMs or managed disks can use Lucidity's storage visibility and autonomous optimization to manage those adjacent block-storage workloads across AWS, Azure, and Google Cloud.
Frequently Asked Questions
Can varchar max be more than 8000?
Yes. varchar(n) allows n from 1 to 8,000 bytes, while varchar(max) holds up to 2^31-1 bytes (2 GB) per value. That's a data-type limit, not the database storage limit.
What is the pricing for Azure SQL Database?
It varies by purchasing model, tier, region, hardware, compute, data and backup storage, redundancy, replicas, and licensing. Use the Azure pricing calculator for current figures.
What is the maximum size of an Azure SQL Database?
It depends on configuration: 2 GB on Basic up to 128 TB on Hyperscale. See the tier table above and verify against Microsoft's limits before deployment.
Can I increase the max size of an Azure SQL Database without changing tiers?
Yes, within the supported MAXSIZE for your current SKU. If you've reached that ceiling, you'll need to scale or migrate.
What happens when an Azure SQL Database reaches its maximum size?
Writes needing growth fail with error 40544 (or 1132 for pools), while SELECT and DELETE still work. Free space or scale, then confirm whether the real limit is data, log, or another resource.