Is your business looking for the best cloud platform that can scale compute and storage independently and handle very large databases (VLDBs)?
Azure SQL Database Hyperscale is a service tier in Microsoft's vCore-based purchasing model. With the largest amount of storage available and the fastest performance possible on Azure, this is one new SQL Server option to keep an eye on.
We talk with many customers searching for on-demand cloud platforms that can support very large databases (VLDB), so our experts have taken the time to explain what it is and how Azure Hyperscale can fit your unique cloud migration goals.
What is Azure SQL Database Hyperscale?
Azure SQL Database Hyperscale is essentially the fastest and most scalable service tier of Azure SQL Database currently available for single database deployments.
As an on-demand service, Hyperscale rapidly adapts to your changing workload needs and auto-scales up to 100 TB for each database. This reduces the need to pre-provision storage resources for big projects, and gives organisations huge potential for application development growth - with no more database storage size limitations.
Azure Hyperscale is specifically optimised for Online Transaction Processing (OLTP) and presently supports all SQL Server workloads, and analytical and hybrid workloads, which means you can migrate your SQL databases to the cloud and take advantage of familiar tools and systems - in addition to the advantages Azure SQL Database offers.
Take note that elastic pools currently don't support the Azure Hyperscale tier, but are will do so in the future.
How does Azure Hyperscale work?
Azure SQL Database Hyperscale decouples compute, log writes and storage to ensure the fastest possible service.
With Hyperscale databases, transactions are processed by writing logs to a log tier with the primary compute node, and fetching data pages from local cache (hot data) or page servers (cooler data) separately. This enables Hyperscale to support the movement of data between different storage tiers (data tiering), leveraging the wide range of storage options available in Microsoft Cloud to get the job done fast.
Any VLDB you have in Hyperscale automatically partitions itself for scale out, and snapshots make database recovery possible within minutes rather than days. Additionally, with the elastic capabilities of Azure, you can scale compute for hyper-scale databases (which are multi-terabyte) up or down almost instantly.
Azure Hyperscale differences
Microsoft has effectively re-architected the current SQL Server storage engine to allow for scale out engine-level compute and storage, so very large databases can be migrated over and supported without scale-up hardware.
Azure Hyperscale differs from other options in Azure which centralise all data management functions in one process by separating the query processing engine from the components, providing long-term storage and durability for data.
Hyperscale also removes several of the practical limits of other cloud databases, as databases are not limited by the resources available in a single node. Databases in Hyper-scale are also not created with a defined maximum size.
Storage capacity with Hyperscale can be scaled out much more seamlessly as needed, with read-only replicas sharing the same compute components, eliminating the need for data copy to spin up new readable replicas.
Azure SQL Database Hyperscale benefits
Hyperscale's biggest advantage is it ensures you're not boxed in by initial configurations - it grows with your needs.
The most significant capability introduced with Azure Hyperscale is you're now able to scale compute and storage resources independently of each other. This enables businesses more experienced with the cloud or those who want to fine-tune and optimise workload performance down to the detail can do so with far more flexibility than before.
Here's Azure Hyperscale's other big business benefits on top of what regular Azure SQL Database service tiers offer:
- Backup: Hyperscale offers near instantaneous database backups, achieved via snapshots of data files taken from page servers at regular intervals to replace traditional streaming backup. Log records are stored in the log service, meaning you can restore your VLDB to any point in time during retention regardless of the size. Finally, because compute and storage are separate, backups don't impact performance of the primary compute node.
Billing: Like all other service tiers, you're only billed for the capacity your business uses (pay for what you use).
- Performance: With Hyperscale, you can leverage faster transaction commit times and higher log throughput commit times, irrespective of the data volumes presently in play - for far greater performance.
- Restore: Based on file snapshots and not a size of data operation, processing restores within the same storage account is much faster then other tiers. Hyperscale also offers overall faster point-in-time recovery.
- Scaling: Hyperscale can instantly scale compute nodes up and down rapidly with shared-storage architecture. One or more read-only nodes can also be provisioned for offloading read workloads or for use as hot-standbys.
- Storage: Automatically scales up to 100 TB per database to help businesses expand their potential for app growth, and makes it easier for a full on-premises database move to Azure SQL PaaS.
Hyperscale stands out from other cloud options in Azure like Managed Instance or Reserved Capacity for its high scalability and performance capabilities that adapts on demand to workload needs - but it's not ideal for everyone.
Who is Azure Hyperscale for?
Hyperscale is designed for enterprises migrating very large databases (VLDB) from on-premises to the cloud. It's also geared toward organisations who want to build new resource-intensive applications on future proof hardware.
Azure Hyperscale is our recommended option for your business if you want to go all-in with your cloud migration, so you can leverage maximum performance and scalability to meet higher-than-average compute and storage needs.
We recommended assessing Hyperscale if you fit into one of three types of customers planning an Azure migration:
- You want to modernise your infrastructure environment: Your business has VLDBs running on-premises that you want to migrate to Azure for a future-proof, highly scalable development platform from the very start.
- You're already in the cloud and need more storage: Your business is currently limited by maximum database size restrictions imposed by other service tiers and want to leverage the auto-scaling storage of Hyperscale.
- You need higher performance and scalability for development: Your organisation has already reached cloud maturity and needs Azure's highest possible compute and storage capabilities to do your business best.
Compared to other options such as Managed Instance which is better for businesses that want to maintain a hybrid environment, Hyperscale fully leverages Azure architecture for maximum scale for both compute and storage resources, and goes far beyond what's available in the General Purpose and Business Critical Service Tiers.
As Azure SQL Database Hyperscale receives more features, we'll be updating this page to keep you in the know - make sure to bookmark us to keep in the loop. In the meantime, visit our Azure Synapse Expertise section.