Troubleshoot and optimize Hyperscale named replicas performance
In the Azure SQL Database Hyperscale service tier, managing named replicas effectively is crucial to maintaining optimal performance. By monitoring log rates and understanding the impact of undersized replicas, you can prevent potential issues such as log rate reduction on the primary replica. Ensuring that each named replica is appropriately sized for its workload can help maintain smooth operations and avoid performance bottlenecks. Remember, the flexibility of Hyperscale allows you to tailor each replica to specific needs, whether through different SLOs or serverless compute options. By staying vigilant and proactive, you can leverage the full potential of Azure SQL Database Hyperscale to meet your workload demands efficiently.Stream data changes from Azure SQL Database - a call for Private Preview of Change Event Streaming
In today's rapidly evolving digital landscape, modern, flexible and real-time data integration from various sources has become more important than ever. Organizations are increasingly relying on data-driven and real-time insights to make informed decisions, optimize operations, and drive innovation. In this context, we are excited to announce a private preview for Change Event Streaming, which enables you to stream your data changes from your SQL database directly into Azure Event Hubs. Starting from today, you can apply for the private preview program. Participants can test the functionality on Azure SQL Database and Azure Event Hubs, and support for more sources and destinations is planned. Participating in the private preview gives you an opportunity to work with the product team, test the functionality, provide your feedback and influence the final release. Note: CES can be tested on Azure SQL Database, and that includes its free offer. Typical use cases for Change Event Streaming are: Building event-driven systems on top of your relational databases, with minimal overhead and easy data integration. Data synchronization across systems, and more specifically, syncing data between microservices or keeping distributed systems in sync. Implementing real-time analytics on top of your relational data. Auditing and monitoring that requires tracking changes of sensitive data or logging specific events. Main advantages for using a message broker such as Azure Event Hubs, and Change Event Streaming are: Scalability, as message brokers are designed to handle high-throughput and can scale independently from a database. Decoupling, as systems downstream from a database and message broker are loosely coupled, enabling greater flexibility and easier maintenance. Multi-consumer support, as Azure Event Hubs allow multiple consumers to process the same data stream, enabling varied use cases from a single source. Real-Time integration, which enables seamless integration between OLTP systems and downstream systems for real-time data flow. If mentioned use cases and advantages resonate with requirements of your architectures, systems and solutions, then Change Event Streaming is the right choice for you. To apply for the private preview, please send email to sqlcesfeedback [at] microsoft [dot] com, and we’ll get back to you with more details! Useful resources Free Azure SQL Database. Free Azure SQL Managed Instance. Azure SQL – Year 2024 in review. Azure SQL YouTube channel.Simplified & lower pricing for Azure SQL Database and Azure SQL Managed Instance backup storage
Today as you deploy your Azure SQL database or Azure SQL managed instance, one of the important decisions to be made is the choice for your backup storage redundancy (BSR). I say it's important because the availability of your database depends on the availability of your backups. Here’s why. Consider the scenario where your DB has high availability configured via zone redundancy. And, let's say, your backups are configured non-zone redundant. In the event of a failure in the zone, your database fails over to another zone within the region, however your backups won't, because of their storage setting. Now, in the new zone, the backup service attempts to backup your database but cannot reach the backups in the zone where the failure happened causing the logs to become full and eventually impacting the availability of the database itself. As you create the Azure SQL database,the choices for backup storage redundancy are: Locally Redundant Storage (LRS) Zone Redundant Storage (ZRS) Geo Redundant Storage (GRS) and Geo Zone Redundant Storage (GZRS) Each of these storage types provides different levels of durability, resiliency and availability for your databases and database backups. Not surprisingly, each storage type also has different levels of pricing, and the price increases significantly as the protection level increases with GZRS storage type almost 4-5x LRS. Choosing between resilience and cost optimization is an extremely difficult choice that the DB owner must make. We are thrilled to announce that, starting from 11/01/2024, the backup storage pricing is now streamlined and simplified across Azure SQL database and Azure SQL Managed Instance. Bonus – we even reduced the prices 😊 The price changes apply to the Backup Storage Redundancy configuration for both Point-in-time and Long-Term Retention backups, across the following tiers of Azure SQL Database and Azure SQL Managed Instance: Product Service Tier Azure SQL Database General Purpose Business Critical Hyperscale Azure SQL Managed Instance General Purpose Business Critical Next Generation General Purpose (preview) As we made the changes, following were the principles we adhered to: No price increase BSR pricing for ZRS is reduced to match the BSR pricing for LRS BSR pricing for GZRS is reduced to match the BSR pricing of GRS BSR pricing for GRS/GZRS will be 2x that of LRS/ZRS Type of backups What is Changing PITR BSR pricing for ZRS is reduced by 20% to match pricing for LRS for all service tiers in Azure SQL Database and Azure SQL Managed Instance except for Azure SQL Database Hyperscale service tier. BSR pricing for GZRS is reduced by 41% to match pricing for GRS for all service tiers in Azure SQL Database and Azure SQL Managed Instance. LTR BSR pricing for ZRS is reduced by 20% to match pricing for LRS for all service tiers in Azure SQL Database and Azure SQL Managed Instance. BSR pricing for GZRS is reduced by 41% to match pricing for GRS for all service tiers in Azure SQL Database and Azure SQL Managed Instance. As an example, lets take East US as the region and look at the pricing for backup storage redundancy for Point in Time storage before and after the changes: For General Purpose/Business Critical service tiers the pricing would now be: Backup Storage Redundancy Current price New Price Price change LRS $0.10 $0.10 None ZRS $0.125 $0.10 20% less GRS $0.20 $0.20 None GZRS $0.34 $0.20 41% less For Hyperscale service tier, the new pricing would now be: Backup Storage Redundancy Current price New Price Price change LRS $0.08 $0.08 None ZRS $0.1 $0.10 None GRS $0.20 $0.20 None GZRS $0.34 $0.20 41% less Similarly, Backup storage redundancy prices for Long Term Retention backups in East US would be as follows: Backup Storage Redundancy Current price New Price Price change LRS $0.025 $0.025 None ZRS $0.0313 $0.025 20% less GRS $0.05 $0.05 None GZRS $0.0845 $0.05 41% less As a customer, the decision now becomes much easier for you. If you need regional resiliency: choose Zone Redundant Storage (ZRS) If you need regional and/or geo resiliency: choose Geo Zone Redundant Storage (GZRS). If the Azure region does not support AvailabilityZones, then choose Local Redundant Storage for regional resiliency, and Geo Redundant Storage for geo resiliency respectively. Please Note: The Azure pricing page and Azure pricing calculator will be updated with these new prices soon. The actual pricing meters have already been updated. Additionally, the LTR pricing change for Hyperscale will be in effect from January 1, 2025.Azure SQL Database Hyperscale – lower, simplified pricing!
REST and GraphQL endpoints, JSON data support, external API invocation. Hyperscale was built leveraging core cloud capabilities and offers auto-scaling, multi-tiered high performance storage, independently scalable compute, read scale-out, predictable, and quick operations like database copy, and much more! We want to ensure that all customers use Hyperscale for their application – no matter what size. Today, we're excited to announce changes to the way Hyperscale is priced. In most cases, you will see significantly lower costs – allowing you to invest the resultant savings in the resources you need to build AI-ready applications, increase the resiliency of your databases, and many other benefits unique to Hyperscale. Let’s take a deeper look at this exciting announcement! What is changing? We are reducing the price of compute by $0.10 USD per vCore per hour (some exceptions are listed later in this post), which in many cases can be up to 35% less than the pre-announcement (“current rate”) compute cost. The storage cost for Hyperscale has also been aligned with the market for developer databases and the pricing for other Azure Database offerings, while not charging for I/O operations. The new pricing will take effect and be displayed on the Azure SQL Database pricing page and Azure pricing calculator on December 15th. Examples of the pricing change Here are some examples to illustrate how the new pricing works as compared to the existing pricing. Note that all costs are estimated and assume a 730 hour month. Case 1: Hyperscale single DB with 6-vCore provisioned compute, 0 HA replica and 50 GB of allocated storage, East US Existing pricing New pricing Compute cost USD 1237.85 USD 800.05 Storage cost USD 5.0 USD 12.5 Total cost USD 1242.9 USD 812.55, saving 35% Case 2: Hyperscale single DB with 12-vCore provisioned compute, 1 HA replica and 200 GB of allocated storage, East US Existing pricing New pricing Compute cost USD 4075.9 USD 3200.20 Storage cost USD 20.0 USD 50.0 Total cost USD 4095.9 USD 3250.20, saving 21% Case 3: Hyperscale single DB with 32-vCore provisioned compute, 1 HA replica and 18 TB of allocated storage, East US Existing pricing New pricing Compute cost USD 10869.08 USD 8533.88 Storage cost USD 1,843.20 USD 4,608.00 Total cost USD 12712.28 USD 13141.88, 3% higher Conclusion In conclusion, these pricing changes for Hyperscale are aligned with our mission to provide the best features, with the highest performance and scalability, at a great price for all our customers. Our team is here to assist with any questions you may have about these changes. Please leave a comment on this blog and we’ll be happy to get back to you. Alternatively, you can also email us at sqlhsfeedback AT microsoft DOT com. We are eager to hear from you all! Frequently Asked Questions When does the change take effect, and what does it impact? The pricing changes take effect on December 15 th , 2023 at 00:00 hours UTC. The changes will apply to the following resources created on or after Dec 15 th , 2023. Any newly created Hyperscale single provisioned compute databases. Any (existing or new) Hyperscale single serverless databases (currently in preview). Any (existing or new) Hyperscale elastic pools (currently in preview). Any newly created Hyperscale elastic pooled databases (currently in preview). What happens to my existing Hyperscale resources? To start with, nothing changes till December 15 th , 2023. Here’s what will happen starting December 15 th , 2023: To provide a seamless experience without any impact on existing workloads, all existing Hyperscale single databases with provisioned compute created before December 15 th , 2023, will continue to be billed at the existing rates, for a period of up to 3 years (ending December 14 th , 2026). Customers will receive further notice of the pricing change to their Hyperscale databases in advance of December 14 th , 2026. All existing Hyperscale single databases with serverless compute (currently in preview) will automatically switch to the new pricing starting December 15 th , 2023. All existing Hyperscale elastic pools (currently in preview) will automatically switch to the new pricing starting December 15 th , 2023. What if I update / change a database to a different deployment option? Hyperscale allows seamless, rapid scaling of the database compute. You can also scale a Hyperscale database to move from provisioned compute to serverless compute (or the other way around). You can add an existing Hyperscale database to an elastic pool or move an elastic pooled database out of the elastic pool to a single database. Here’s how your costs are impacted if you perform any of these changes on or after December 15 th , 2023: Change Impact Hyperscale (serverless single, or elastic pooled) database is changed to a Hyperscale single database with provisioned compute. The final cost of the database will be based on when the database was created. If the database was created prior to December 15 th , 2023, it will be billed as per the existing pricing. If the database was created on or after December 15 th , 2023, it will be billed as per the new pricing. Hyperscale database is changed to a Hyperscale single database with serverless compute. The database will be billed with the new pricing. Hyperscale database is added to an elastic pool on or after December 15 th. The database’s storage will be charged as per the “new” storage pricing. There is no separate compute cost for a database in an elastic pool. Hyperscale single database with provisioned compute is scaled up, or down, or has high-availability replicas added or removed, or its hardware family is changed. The pricing model remains as it was before the scaling operation. The actual costs for compute resources will change based on the scaling operation (for example, they will increase if the database is scaled up, or replicas added). Do note that if you are using reserved capacity ("reservations") and are changing the hardware family, you will need to exchange those reservations to align to the new hardware family. The costs of storage resources associated with the single database will not change due to the scaling operation itself. Any copies of a Hyperscale database created as a Hyperscale single database, on or after December 15 th , 2023. The database copies will use the new pricing, regardless of when the original database was created. Any new single database created via. restore, or geo-replication operations, on or after December 15 th , 2023. The new database will use the new pricing, regardless of when the original database was created. Any non-Hyperscale database is updated to Hyperscale on or after December 15 th , 2023. The new database will use the new pricing, regardless of when it was originally created. See the summarized tables below for a quick reference. Single databases Hyperscale single databases with provisioned compute Hyperscale single databases with serverless compute Timeline: before December 15 th , 2023 Timeline: on or after December 15 th , 2023 Timeline: before December 15 th , 2023 Timeline: on or after December 15 th , 2023 Database was created or migrated to Hyperscale before December 15th, 2023 Database created or migrated to Hyperscale after December 15th, 2023 Database was created or migrated to Hyperscale before December 15th, 2023 Database created or migrated to Hyperscale after December 15th, 2023 Compute Existing provisioned compute price. Existing provisioned compute price. New provisioned compute price. Existing serverless compute price. New serverless compute price. New serverless compute price. Storage Existing storage prices. Existing storage prices. New storage prices. Existing storage prices. New storage prices. New storage prices. Elastic pools and pooled databases Hyperscale elastic pools Hyperscale elastic pooled databases Timeline: before December 15 th , 2023 Timeline: on or after December 15 th , 2023 Timeline: before December 15 th , 2023 Timeline: on or after December 15 th , 2023 Elastic pool was created before December 15th, 2023 Elastic pool created after December 15th, 2023 Database was created or migrated to Hyperscale before December 15th, 2023 Database created or migrated to Hyperscale after December 15th, 2023 Compute Existing provisioned compute price. New provisioned compute price. New provisioned compute price. N/A (charged per elastic pool) Storage N/A – storage is charged per database. Existing storage prices. New storage prices. New storage prices. Can I continue to use reservations for Hyperscale? With reservations, you make a commitment to use SQL Database for a period of one or three years to get a significant discount on the compute (vCores) costs. There are no changes to the Compute (vCores) pricing, and you can continue to use reserved capacity (“reservations”) for Hyperscale single databases with provisioned compute and Hyperscale elastic pools. How can I move my existing Hyperscale databases to the new pricing? Currently, there is no in-built support to switch pricing for existing Hyperscale databases. However, you can consider one of the redeployment methods (database copy, point-in-time restore, or geo-replication) to create a new “copy” of the existing Hyperscale database. The newly created “copy” of the database will be billed the new pricing. If you do decide to go down this path, do consider creating the new database with zone redundancy, as described here. Do you have any projections of likely costs when converting non-Hyperscale DBs to Hyperscale? We recommend you use the Azure pricing calculator to compare base cost for compute and storage. However, costs of backups can vary depending on the nature of the workload and backup retention settings configured. Databases in the (DTU) Basic, Standard and Premium service tiers include backup in the base cost. When converting such databases to Hyperscale, keep in mind that backups in Hyperscale can be a significant factor in overall cost. It is only possible to determine this after sufficient testing with realistic workloads on Hyperscale and we strongly recommend you do such testing before converting DTU service tier databases to Hyperscale. Does the reduction in compute price apply to all subscription offer types? In the case of dev/test subscriptions and related offer types including Enterprise dev/test, where you were already not paying for license costs, there will not be a further reduction in the price of compute. For such subscriptions, the storage costs for Hyperscale resources will still be based on the guidelines in the “When does the change take effect, and what does it impact?” and “What happens to my existing Hyperscale resources” sections in this blog post. Can I still use Azure Hybrid Benefit for Hyperscale? Azure Hybrid Benefit no longer applies to the Hyperscale tier, except forHyperscale single databases with provisioned compute which were created prior to December 15, 2023. Even for those older databases, Azure Hybrid Benefit can only be used till December 14, 2026. Note that specifying values of BasePrice or LicenseIncluded for the LicenseType parameter in APIs / SDKs / PowerShell / CLI, is only relevant for Hyperscale single databases with provisioned compute which were created prior to December 15, 2023. These values are effectively ignored for all other types of Hyperscale resources. Current limitations and known issues Consider a database which was created originally as a non-Hyperscale database prior to Dec 15th, 2023. If this database is then migrated to Hyperscale on or after Dec 15th, 2023, the Cost Summary section of the Azure portal will incorrectly show the "old pricing" for this database. The Cost Summary section in the Azure portal is only intended to be an estimate. We recommend you rely on the Cost Management section in the Azure portal to review actual costs. Azure portal cost summary view in greater China regions does not show the updated pricing information accurately. This is only a display issue and does not impact the billing in any way. Please refer to the pricing calculator or thepricing page for accurate pricing information.Announcing enhancements to Azure SQL Database Hyperscale
We are thrilled to announce a set of significant updates to Azure SQL Database Hyperscale. These updates are designed to provide more scalability, faster data processing, and improved reliability, particularly during failovers Increased maximum database size We are pleased to announce that the maximum database size in Azure SQL Database Hyperscale has increased from 100 TiB to 128 TiB. This enhancement is now generally available (GA) for single Hyperscale databases and will be released later for Hyperscale elastic pools. This expansion allows for even greater flexibility and capacity to manage large datasets, accommodating the needs of businesses with substantial data storage requirements. Higher limits for transaction log generation rate In our continuous effort to improve performance, the log generation rate in Azure SQL Database Hyperscale has been increased from 100 MiB/s to 150 MiB/s. The increased limit for the transaction log generation rate limit is currently in limited public preview, and you can sign up using this form: link. The higher log generation rate means faster data processing and better handling of write-intensive workloads. This ensures that your applications run smoothly and efficiently, even during peak usage times. Whether you're dealing with bulk data inserts, high-volume transaction processing, real-time data ingestion, or rebuilding of large indexes, the enhanced log generation rate provides the performance boost needed to keep your systems responsive and reliable. Continuous priming One another new feature we are introducing is continuous priming. This innovative feature is designed to optimize performance during failovers by priming secondary compute replicas. Here’s how it works: Continuous priming collects information about the most frequently accessed pages in all Hyperscale compute replicas, both primary and secondary. This information is aggregated at the Hyperscale storage layer (page servers). All Hyperscale compute replicas then use this list of most frequently accessed pages, which represents the typical customer workload, to “prime” both the buffer pool (BP) and the resilient buffer pool extension (RBPEX) with any missing pages. This “priming” process is done continuously to keep up with changes in the customer work set. With continuous priming, local HA replicas will prime themselves with the pages being used in the primary replica. This ensures that performance remains consistent and optimized, even during failovers. Please note that continuous priming is not applicable to Hyperscale databases with serverless compute and named replicas. If you’re interested in enrolling in this preview, please sign up using the provided link. Conclusion We are confident that these new features and enhancements will significantly benefit your operations, providing more scalability, faster data processing, and greater reliability. Stay tuned for more updates as we continue to innovate and improve Azure Database Hyperscale to meet your needs. Please share your feedback and questions by leaving a comment; you can also email us at sqlhsfeedback AT microsoft DOT com. We are eager to hear from you all!Elastic pools for Azure SQL Database Hyperscale now in preview
Elastic pools for Azure SQL Database Hyperscale now in preview!Azure SQL Database Hyperscale elastic pools (“Hyperscale elastic pools”) combines the leading-edge architecture for Hyperscale with the cost effectiveness of elastic pools by introducing a shared resource model for Hyperscale databases.
