Microsoft is taking the mystery out of performance issues that may impact the performance of SQL database workloads on the company’s Azure cloud.
The Query Store service for Azure SQL is now on and available by default, the company has announced. “Query Store acts as a ‘flight data recorder’ for the database, continuously collecting critical information about the queries,” wrote Borko Novakovic, a senior program manager with Microsoft’s SQL Server group. “It dramatically reduces resolution time in case of performance incidents, as pre-collected, relevant data is available when you need it, without delays.”
With Query Store, users can determine the impact of an application or configuration change on workload performance by using the feature’s A/B testing capabilities. It can also be used to identify and tune resource-intensive queries by monitoring CPU time, read/write operations, elapsed time and other data consumption metrics.
Customers can turn off Query Store, but the benefits outweigh the potential downsides. Novakovic recommends that users leave the service on, asserting that it has what he described as negligible impact of 1 or 2 percent on performance overhead.
Meanwhile, customers using Azure SQL Database elastic pools have several new subscription options to consider. The company has added new smaller elastic pool sizes, measured by eDTUs (elastic database transaction units), and has rolled out new pricing that lowers the cost of moving up or down pools sizes.
Sizes range from 50 eDTUs (Basic and Standard) to 4,000 eDTUs (Premium). An eDTU describes a set of resources (CPU, memory, data I/O and transaction log I/O) that can be shared between databases.
Standard Azure SQL elastic pools now support up to 2.9 TB of storage (3,000 eDTUs). Per-pool database limits have been raised to 100 databases for Premium pools of (250 eDTUs minimum) and 500 databases for Basic and Standard pools (200 eDTUs minimum). Finally, customers with large Premium pools can now accommodate databases with up to 1,750 eDTUs (P11 performance level) and 4,000 eDTUs (P15 performance level).
Building on Microsoft Research’s work on speech recognition, the company also announced Japanese language support for Azure Media Indexer v.2, allowing customers to generate transcripts from their media files. The company also lifted the 10-minute processing limit and made improvements that affect the cloud service’s punctuation and grammar capabilities.
Microsoft also announced that its Azure Redis Cache solution, the company’s open-source data caching platform, is being moved to the upcoming Azure Monitor service’s infrastructure. This will enable the company to improve on Azure Redis Cache’s monitoring and alerting capabilities.
After the upgrade, users will no longer have to create storage accounts and configure the solution to collect metrics, promised Ye Gu, principal program manager of Microsoft Azure Developer Experience.
“All caches will automatically display metrics in the Azure service portal,” he blogged. “These new Redis metrics data also will be accessible through Azure Monitor service’s REST API. If you have a need to retain metrics for longer than 30 days, you may still export it to your own storage account for archiving and offline analysis.”