Build highly scalable, AI-ready applications on Azure SQL Database Hyperscale

Asad Khan

February 14th, 20240 5

Build highly scalable, AI-ready applications on Azure SQL Database Hyperscale

AI and cloud-native applications built for the future need a highly scalable, cloud native database with excellent price-performance. For applications operating on mission-critical relational data, Azure SQL Database Hyperscale provides the performance, reliability and security you expect from Azure SQL – but at the same price as commercial open source databases.

SQL Database Hyperscale delivers consistently high performance at any scale, standing above the competition. Combined with a new lower price, choosing the right relational database for your next application just got easier in a market full of relational database management systems (RDBMSs).

Faster is better

You can build your next app on many databases, including Amazon Aurora, but you’ll want to take another look at Azure SQL Database Hyperscale. We partnered with Principled Technologies on a commissioned study to compare Azure SQL Database Hyperscale performance with other cloud databases in the market. Principled Technologies’ study clearly shows that Azure SQL Database Hyperscale scales to manage your most critical workloads, outpacing Amazon Aurora PostgreSQL by up to 68 percent in both performance and value.1

A screenshot of a graph Description automatically generated

Azure SQL Database Hyperscale outperforms Amazon Aurora in benchmark tests performed by Principled Technologies.

The benchmark test used the HammerDB 4.8 open-source OLTP workload (TPROC-C), a widely recognized industry standard. Using the ‘all warehouses‘ option, Principled Technologies performed three tests on IO intensive workloads with three different user counts of 64, 96, and 128 users, and a database size of 2.3TB. Azure SQL Database Hyperscale processed new orders much faster than Amazon Aurora for the same vCore configurations, even as transaction volume increased.

Intelligent autoscaling for the coolest apps you can dream up

If you’re a developer, you know the challenges of managing data growth, performance, and scalability for your applications. You need a database that can deliver fast query responses, scale up or down on demand and handle massive amounts of data – particularly when training AI models. But just how big of a database do you need? When starting out, you may not know.

One may think that hyperscale means hypersized. It’s better to think of SQL Database Hyperscale as the hyperflexible solution designed for today’s data-intensive innovations of any size, including the latest AI-ready SQL apps. With its flexible storage architecture, SQL Database Hyperscale grows as needed for your ever more intelligent apps. In fact, Hyperscale databases are not created with a defined max size, storage is allocated automatically—and you are billed only for the storage capacity actually used. Regardless of the type of workload, you can take advantage of the high throughput, autoscaling storage, and fast backups, for improved processing times and price-performance efficiency. SQL Database Hyperscale is the solution that keeps up as your apps grow to serve millions of users.

That’s why performing rights nonprofit Broadcast Music, Inc. (BMI) is using SQL Database Hyperscale to process ever-growing terabytes of data more efficiently and affordably.

“We’ve improved our processing times by a minimum of 20 percent and as much as several hundred percent in some system areas.” Greg Walrath, BMI Executive Director of Enterprise Architecture

Price-performance value is also why one of the world’s largest logistics companies – DHL – use SQL Database Hyperscale to handle its projected growth. According to Jochen Fischer, Deutsche Post DHL Group Senior Software and Security Architect, “The primary motivation was the price tag for Azure. We found that it would save us a lot of money.”

Build AI-ready apps on Azure SQL Database

Artificial intelligence (AI) is transforming the way businesses operate, innovate, and compete. Today, you can harness the power of AI through the close integration that Azure SQL Database has with the broader suite of AI-related services on Azure. Integration with Azure Open AI and Azure AI Search is powering new scenarios like vector search and Retrieval Augmentation Generation (RAG) for working with large language models (LLMs). For example, you can create Azure OpenAI service models for your applications and query your data more easily. You can also use SQL Database Hyperscale together with Azure AI Search to train LLMs on your own data.

Market Leading Database Capabilities

Azure SQL Database Hyperscale brings three decades of SQL innovation to developers, along with the cloud scale and the latest AI capabilities. Some salient features that differentiate SQL Hyperscale database layer from the competition include:

  • Storage snapshot capability in Hyperscale provides fast backup and restore, that meets the aggressive requirements of cloud native workloads. Support for 30 Named Replicas allowing different compute sizes than the primary for each replica is not offered in any other cloud database service. This is a huge advantage for highly transactional and read-scalable workloads both in terms of performance as well as cost.
  • Named replicas in conjunction with the support for three types of secondary replicas i.e. Read Scale-Out, HA and Geo Replication make Hyperscale even more unique and differentiated.
  • Rapid & predictable scale up/down capability – Uniquely helps developers meet aggressive operations and sustain SLA requirements for highly scalable very large database (VLDB) systems.
  • Independent of database size, the Premium & Memory Optimized SKUs of Hyperscale uniquely positions it to better handle higher throughput & lower latency (using RBPEX caching) transactional workloads i.e., Hyperscale database can handle 300K+ IOPS with 128 vCores.
  • Growing trend of ISVs wanting to leverage SQL Database Hyperscale elastic pool capability that allows them to package many databases into an elastic pool, enabling them to optimize price-performance and database operations.

Get started on Azure SQL Database today

Power the application you want to build with just a few clicks. Try Azure SQL Database free of charge and get 100,000 vCore seconds of general purpose serverless compute each month for the lifetime of your subscription.

Learn more

For more information:

1 Price-performance claims based on data from a study commissioned by Microsoft and conducted by Principled Technologies in December 2023. The study compared performance and price performance between a 16 vCore and 32 vCore Azure SQL Database using premium-series hardware on the Hyperscale service tier and the db.r6i.4xlarge and db.r6i.8xlarge offerings for Amazon Aurora PostgreSQL I/O-Optimized (“Amazon Aurora”). Benchmark data is taken from a Principled Technologies report which used the HammerDB TPROC-C benchmark. The TPROC-C workload is derived from the TPC-C Benchmark and results were obtained with the HammerDB TPROC-C workload. The HammerDB TPROC-C workload is derived from the TPC-C benchmark and is not comparable to published TPC-C Benchmark results, as this implementation does not comply with all requirements of the TPC Benchmark. Price-performance is calculated by Principled Technologies as the cost of running the cloud platform continuously divided by new orders per minute throughput, based upon the standard. Prices are based on publicly available US pricing in East US 1 for Azure SQL Database and US East for Amazon Aurora as of December 2023. Performance and price-performance results are based upon the configurations detailed in the Principled Technologies report. Actual results and prices may vary based on configuration and region.

Asad Khan Vice President SQL, SQL

Read next

0 comments

Discussion is closed.

Feedback usabilla icon