March 2023 Recap: Azure Database for PostgreSQL Flexible Server

[Guest kabharati]

We are pleased to present our March 2023 update, which comes with an array of exciting news and enhancements in Azure Database for PostgreSQL Flexible Server. This month, we have achieved a significant milestone by announcing the Single Server Retirement and the general availability of the Single Server to Flexible Server Migration tool, which simplifies the migration of Single Server workloads to Flexible Server. We announced the much-anticipated general availability of Read Replicas and New Burstable SKUs (B4ms- B20ms), which provides a cost-effective solution for flexible CPU usage to support workloads with varying usage patterns. Additionally, we have announced the general availability of Azure Monitor workbooks, preview of PgBouncer metrics and launched Flexible Server in Australia Central region. Lastly, we have made extensive updates to our product documentation to bring the latest information to our customers.

 

 

 

We invite you to keep reading for more information on how these updates can benefit you and your organization.

 

 

 

 

 

• Single Server Retirement Announcement
  

 

 

 

• Single Server to Flexible Server Migration Tool (Generally Available)
  

 

 

 

• Read Replica Support (Generally Available)
  

 

 

 

• New Burstable SKUs (B4ms -B20ms) (Generally Available)
  

 

 

 

• Azure Monitor Workbooks (Generally Available)
  

 

 

• PgBouncer metrics (Preview)
  

 

 

 

Single Server Retirement & Single Server to Flexible Server Migration Tool (Generally Available)

 

 

 

 

On 3/30, we announced that Azure Database for PostgreSQL – Single Server will be retired on March 28 2025. We will continue to support Single Server through March’2025, however, we strongly recommend that you consider migrating to Flexible Server, the next generation deployment option in Azure Database for PostgreSQL. Our decision to retire Azure Database for PostgreSQL - Single Server in 2025 marks a shift towards our investment and vision for a more robust and scalable  Azure Database for PostgreSQL - Flexible Server.

 

 

 

We will provide support and maintenance for existing Single Server workloads until it is retired in 2025 and will work closely with our customers to ensure a smooth migration to Flexible Server by minimizing any potential disruptions. To help with this transition, we have also announced the general availability of the of Single Server to Flexible Server Migration tool and additional migration guidance to consider when planning this move. Currently, the Single Server to Flexible Server Migration tool only supports offline migrations, but we plan to introduce support for online migrations soon. The migration tool leverages a new binary named pgcopydb, which enables rapid and efficient copying of databases from one server to another. Despite relying on traditional methods like pg_dump and pg_restore for schema migration, pgcopydb employs its own data migration mechanism that streams parts from source to target using multiple processes. Moreover, pgcopydb operates differently than pg_restore in terms of index building, as it constructs all indexes simultaneously, resulting in a swifter data migration process.

 

 

 

Here is how you can initiate a single server migration using the tool from a flexible server. In the Overview tab of the Flexible Server select Migration and click Migrate from Single Server.

 

 

 

 

 

 

 

Learn more about Migration Strategies from Single to Flexible Server

 

 

 

Please go through this blog for more information on single server retirement announcement.

 

 

 

Read Replicas (Generally Available)

 

 

 

 

Azure Database for PostgreSQL – Flexible Server now supports read replicas, a feature that enables data replication from a primary server to a read-only replica asynchronously using PostgreSQL's native physical replication technology. This feature helps improve the performance and scalability of read-intensive workloads by isolating read workloads to the replicas while directing write workloads to the primary server. Additionally, read replicas can be in any global region and can be promoted to a read-write server in the event of a disaster recovery. Replicas are new servers that you manage like regular Azure Database for PostgreSQL flexible servers. For each read replica, you'’re billed for the provisioned compute in vCores and storage in GB/ month. You can freely change most of the server parameters on your read replica server, allowing you to set different values than on the primary server.

 

 

 

The read replica feature supports scenarios where some delay is acceptable, and read workloads can be offloaded, rather than synchronous replication, where the replica data should be up to date. The time delay between the primary server and replica can vary from minutes to hours depending on the workload and latency between the primary and the replica, therefore the read replica feature should be used for workloads that can tolerate this potential delay like DR scenarios.

 

 

 

Flexible Server supports both In-Region and Cross Region replicas. In-region replicas uses snapshot to create replicas whereas cross-region or geo replicas uses base backup of the primary instance, which is then transmitted over the network therefore duration of replica creation might range from minutes to several hours depending on the primary size.

 

 

 

Learn more about read replicas from the GA announcement and the documentation.

 

 

 

To configure replicas please select the Replication Tab on Flexible Server and click Add Replicas as shown in below screenshot.

 

 

 

 

 

 

 

 

 

 

New Burstable SKUs (Generally Available)

 

 

 

 

Azure Database for PostgreSQL-Flexible Server now supports five new burstable SKUs (B4ms, B8ms, B12ms, B16ms, B20ms), which enable customers to utilize flexible CPU usage and accommodate workloads with varying usage patterns. These SKUs are particularly beneficial for small database workloads that require minimal CPU performance for extended periods but may require full CPU power due to incoming traffic or workload requirements. With up to 20 vCPUs and 80 GiB of RAM, these SKUs provide a cost-effective deployment solution for workloads that experience periodic bursts in performance rather than continuous maximum CPU performance, helping customers save costs while meeting the demands of unpredictable workloads.

 

 

 

 

 

Learn more about new Burstable SKUs from the documentation.

 

 

 

 

PgBouncer metrics (Preview)

 

 

 

 

 

 

Azure Database for PostgreSQL – Flexible Server has a built-in option for customers to easily enable PgBouncer, that is widely used with PostgreSQL for connection pooling. By enabling PgBouncer customers can update their app connection string to talk to PgBouncer on port 6432 instead of directly to the database server. PgBouncer implements the Postgres wire protocol so that your app will still “think” it is connecting to the Postgres database server. Monitoring PgBouncer can be difficult due to its use of SHOW commands that are only available through a special database connection. To address this issue, we have released six new metrics that allow customers to monitor PgBouncer connection pooling performance in Azure Database for PostgreSQL Flexible Server. These metrics emit every 30 minutes and have up to 93 days of retention, allowing customers to configure alerts and access new metrics dimensions for filtering and splitting data on the database name. Monitoring these metrics can help customers gain insights into their connection pool performance and identify any issues before they become critical.

 

 

 

Learn more about PgBouncer Metrics from the documentation and from the blog announcement.

 

 

 

Azure Monitor Workbooks (Generally Available)

 

 

 

 

Azure Monitor workbooks are now generally available in Azure Database for PostgreSQL Flexible Server, providing users with a flexible canvas to analyze data and create visually rich reports within the Azure portal. Workbooks allow users to access multiple data sources across Azure and combine them into a unified interactive experience. Workbook templates are pre-designed reports that can be reused by different users and teams.

 

 

 

Upon opening a template, a transient workbook is created, which is populated with the contents of the template. With this integration, users can link to workbooks and sample templates, which can help monitor the service at scale. Users can edit these templates, customize them to their specific requirements, and pin them to the dashboard to create a focused and organized view of Azure resources.

 

 

 

Monitor workbooks for Azure Database for PostgreSQL Flexible Server

 

 

 

Other Notable Releases

 

 

 

 

Azure Database for PostgreSQL Flexible Server now supports the Australia Central region.

 

 

 

Finally, we have made significant updates to our product documentation pages to ensure that our customers are well-informed with the latest information and product releases. Additionally, we also added the tag below to all our Single Server documentation to notify our customers about single server retirement plans.

 

 

 

 

 

 

 

 

 

 

Learn more about  Azure Database for PostgreSQL - Flexible Server.

 

 

That is all for this month! Please continue sending us your feedback, please reach out via email to Ask Azure DB for PostgreSQL for any other questions or clarifications.

 

Continue reading...