Merge pull request #36297 from MicrosoftDocs/main

Auto Publish – main to live - 2026-01-19 23:30 UTC

Author: learn-build-service-prod[bot]

docs/database-engine/availability-groups/windows/database-snapshots-with-always-on-availability-groups-sql-server.md

@@ -1,9 +1,10 @@
 ---
-title: "Create a database snapshot for an availability group"
+title: "Create a Database Snapshot for an Availability Group"
 description: "Describes how to create a database snapshot for a database within an Always On availability group on either the primary or secondary database."
 author: MashaMSFT
 ms.author: mathoma
-ms.date: "05/17/2016"
+ms.reviewer: randolphwest
+ms.date: 01/19/2026
 ms.service: sql
 ms.subservice: availability-groups
 ms.topic: how-to
@@ -12,27 +13,30 @@ helpviewer_keywords:
   - "database snapshots [SQL Server], Always On Availability Groups"
   - "Availability Groups [SQL Server], interoperability"
 ---
-# Database Snapshots with Always On Availability Groups (SQL Server)
+# Database snapshots with Always On availability groups (SQL Server)
+
 [!INCLUDE [SQL Server](../../../includes/applies-to-version/sqlserver.md)]
 
-[!NOTE] Creating database snapshots on any database has cpu and IO overhead due to copy on write activity. On database replicas, this can hurt redo throughput among other operations, especially as the number of snapshots increases.
-
-  You can create a database snapshot on an primary or secondary database in an availability group. The replica role must be either PRIMARY or SECONDARY, not in the RESOLVING state.  
-  
- We recommend that the database synchronization state be SYNCHRONIZING or SYNCHRONIZED when you create a database snapshot. However, database snapshots can be created when the database synchronization state is NOT SYNCHRONIZING.  
-  
- A database snapshot on a secondary replica should continue to work if the replica is DISCONNECTED from the primary replica.  
-  
- Some [!INCLUDE[ssHADR](../../../includes/sshadr-md.md)] conditions cause both the source database and its database snapshots to be restarted, temporarily disconnecting users. These conditions are as follows:  
-  
--   The primary replica changes roles, whether because the current primary replica goes off line and comes back online on the same server instance or because the availability group fails over.  
-  
--   The database enters the secondary role.  
-  
- If the availability replica that hosts database snapshots is failed over, the database snapshots remain on the server instance where they were created. Users can continue to use the snapshots after the failover.If performance is a concern in your environment, we recommend that you create database snapshots only on secondary databases that are hosted by a secondary replica that is configured for manual failover mode.  If you ever manually fail over the availability group to this secondary replica, you can create a new set of database snapshots on another secondary replica, redirect clients to the new database snapshots, and drop all of the database snapshots from the now primary databases.  
-  
-## See Also  
- [Overview of Always On Availability Groups (SQL Server)](../../../database-engine/availability-groups/windows/overview-of-always-on-availability-groups-sql-server.md)   
- [Database Snapshots (SQL Server)](../../../relational-databases/databases/database-snapshots-sql-server.md)  
-  
-  
+You can create a database snapshot on a primary or secondary database in an availability group. The replica role must be either `PRIMARY` or `SECONDARY`, and can't be in the `RESOLVING` state.
+
+> [!NOTE]  
+> Creating database snapshots on any database adds CPU and I/O overhead because of copy-on-write activity. On database replicas, this overhead can reduce redo throughput and affect other operations, especially as the number of snapshots increases.
+
+You should create database snapshots when the database synchronization state is `SYNCHRONIZING` or `SYNCHRONIZED`. However, you can still create database snapshots when the database synchronization state is `NOT SYNCHRONIZING`.
+
+A database snapshot on a secondary replica continues to work if the replica is `DISCONNECTED` from the primary replica.
+
+Some [!INCLUDE [ssHADR](../../../includes/sshadr-md.md)] conditions cause both the source database and its database snapshots to restart, temporarily disconnecting users. These conditions are as follows:
+
+- The primary replica changes roles. This change can happen because the current primary replica goes offline and comes back online on the same server instance, or because the availability group fails over.
+
+- The database enters the secondary role.
+
+If the availability replica that hosts database snapshots fails over, the database snapshots remain on the server instance where you created them. You can continue to use the snapshots after the failover. If performance is a concern in your environment, create database snapshots only on secondary databases hosted by a secondary replica that is configured for manual failover mode.
+
+If you ever manually fail over the availability group to this secondary replica, you can create a new set of database snapshots on another secondary replica, redirect clients to the new database snapshots, and drop all of the database snapshots from the now primary databases.
+
+## Related content
+
+- [What is an Always On availability group?](overview-of-always-on-availability-groups-sql-server.md)
+- [Database snapshots (SQL Server)](../../../relational-databases/databases/database-snapshots-sql-server.md)

docs/relational-databases/system-dynamic-management-views/linux-related-dynamic-management-views-and-functions-transact-sql.md

@@ -17,15 +17,16 @@ dev_langs:
 
 [!INCLUDE [sqlserver2025-linux](../../includes/applies-to-version/sqlserver2025-linux.md)]
 
-The following dynamic management objects return system-level Linux metrics beyond SQL Server, for a comprehensive visibility into overall host performance.
+The following dynamic management objects return system-level Linux metrics beyond SQL Server, so you get comprehensive visibility into overall host performance.
 
-These objects provide detailed CPU statistics, real-time network interface activity (bytes and packets sent and received, errors, drops, and collisions), and disk I/O statistics for each storage device.
+These objects provide detailed CPU statistics, real-time network interface activity (bytes and packets sent and received, errors, drops, and collisions), disk I/O statistics for each storage device, and virtual memory statistics.
 
 ## In this section
 
 - [sys.dm_os_linux_cpu_stats](sys-dm-os-linux-cpu-stats-transact-sql.md)
 - [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md)
 - [sys.dm_os_linux_net_stats](sys-dm-os-linux-net-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats](sys-dm-os-linux-vm-stats-transact-sql.md)
 
 ## Related content
 

docs/relational-databases/system-dynamic-management-views/sys-dm-os-linux-cpu-stats-transact-sql.md

@@ -53,23 +53,27 @@ Requires `VIEW SERVER PERFORMANCE STATE` permission on the server.
 
 ## Remarks
 
-This DMV provides cumulative statistics since the system was started.
+This DMV provides cumulative statistics since the system started.
 
-Values aren't limited to SQL Server activity; they reflect the entire Linux host.
+Values aren't limited to SQL Server activity. They reflect the entire Linux host.
 
 For more information on interpreting Linux CPU statistics, see the `proc_stat(5)` manual page for your distribution.
 
-Use this DMV with other Linux-specific DMVs, such as [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md) and [sys.dm_os_linux_net_stats](sys-dm-os-linux-net-stats-transact-sql.md), for holistic monitoring.
+Use this DMV with other Linux-specific DMVs for holistic monitoring:
+
+- [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md)
+- [sys.dm_os_linux_net_stats](sys-dm-os-linux-net-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats](sys-dm-os-linux-vm-stats-transact-sql.md)
 
 ### Usage scenarios
 
 - **Diagnose CPU saturation**: Use `loadavg_1min`, `proc_runable_cnt`, and `user_time_cs` to identify if the system is under heavy CPU load.
 
 - **Investigate I/O waits**: High `iowait_time_cs` or `proc_ioblocked_cnt` values could indicate storage bottlenecks.
 
-- **Analyze system responsiveness**: Frequent context switches (`csw_cnt`) or interrupts (`interrupt_cnt`) can signal issues with system scheduling or hardware.
+- **Analyze system responsiveness**: Frequent context switches (`csw_cnt`) or interrupts (`interrupt_cnt`) can signal problems with system scheduling or hardware.
 
-- **Correlate SQL Server performance with system activity**: Since these metrics are system-wide, they help distinguish between SQL Server-specific and broader OS-level issues.
+- **Correlate SQL Server performance with system activity**: Since these metrics are system-wide, they help you distinguish between SQL Server-specific and broader OS-level problems.
 
 ## Examples
 
@@ -128,4 +132,5 @@ WHERE scheduler_id < 255;
 
 - [sys.dm_os_linux_disk_stats (Transact-SQL)](sys-dm-os-linux-disk-stats-transact-sql.md)
 - [sys.dm_os_linux_net_stats (Transact-SQL)](sys-dm-os-linux-net-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats (Transact-SQL)](sys-dm-os-linux-vm-stats-transact-sql.md)
 - [Performance best practices and configuration guidelines for SQL Server on Linux](../../linux/sql-server-linux-performance-best-practices.md)

docs/relational-databases/system-dynamic-management-views/sys-dm-os-linux-disk-stats-transact-sql.md

@@ -64,7 +64,11 @@ Requires `VIEW SERVER PERFORMANCE STATE` permission on the server.
 
 - **Use with care in multitenant hosts**: High values can originate from non-SQL Server workloads. Correlate with system tools or other DMVs for attribution.
 
-Use this DMV with other Linux-specific DMVs, such as [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md) and [sys.dm_os_linux_net_stats](sys-dm-os-linux-net-stats-transact-sql.md), for holistic monitoring.
+Use this DMV with other Linux-specific DMVs for holistic monitoring:
+
+- [sys.dm_os_linux_cpu_stats](sys-dm-os-linux-cpu-stats-transact-sql.md)
+- [sys.dm_os_linux_net_stats](sys-dm-os-linux-net-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats](sys-dm-os-linux-vm-stats-transact-sql.md)
 
 ### Usage scenarios
 
@@ -203,4 +207,5 @@ ORDER BY ios_in_progress DESC,
 
 - [sys.dm_os_linux_cpu_stats (Transact-SQL)](sys-dm-os-linux-cpu-stats-transact-sql.md)
 - [sys.dm_os_linux_net_stats (Transact-SQL)](sys-dm-os-linux-net-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats (Transact-SQL)](sys-dm-os-linux-vm-stats-transact-sql.md)
 - [Performance best practices and configuration guidelines for SQL Server on Linux](../../linux/sql-server-linux-performance-best-practices.md)

docs/relational-databases/system-dynamic-management-views/sys-dm-os-linux-net-stats-transact-sql.md

@@ -53,7 +53,11 @@ This DMV provides system-level network metrics and isn't limited to SQL Server a
 
 `sys.dm_os_linux_net_stats` is useful for diagnosing network problems that might affect SQL Server connectivity, Always On availability groups, replication, or client/server communication.
 
-Use this DMV with other Linux-specific DMVs, such as [sys.dm_os_linux_cpu_stats](sys-dm-os-linux-cpu-stats-transact-sql.md) and [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md), for holistic monitoring.
+Use this DMV with other Linux-specific DMVs for holistic monitoring:
+
+- [sys.dm_os_linux_cpu_stats](sys-dm-os-linux-cpu-stats-transact-sql.md)
+- [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats](sys-dm-os-linux-vm-stats-transact-sql.md)
 
 ### Usage scenarios
 
@@ -80,7 +84,7 @@ FROM sys.dm_os_linux_net_stats;
 
 ### B. Identify interfaces with packet errors or drops
 
-Use the following query to focus on interfaces experiencing problems, filtering for nonzero error or drop counts.
+Use the following query to focus on interfaces experiencing problems. It filters for nonzero error or drop counts.
 
 ```sql
 SELECT interface,
@@ -99,7 +103,7 @@ WHERE recv_errors > 0
 
 Measure network throughput for each interface over a specific interval. This measurement helps you in capacity planning, and identifies bottlenecks during peak workloads.
 
-The following query tracks network usage trends, periodically samples bytes sent and received, and calculates the delta.
+The following query tracks network usage trends. It periodically samples bytes sent and received, and calculates the delta.
 
 Take a baseline snapshot:
 
@@ -135,7 +139,7 @@ WHERE recv_multicast > 0;
 
 ### E. Correlate network statistics with wait statistics
 
-If you observe high `NETWORK_IO` waits, check for corresponding network errors or drops to pinpoint the root cause of query delays.
+If you see high `NETWORK_IO` waits, check for matching network errors or drops to identify the root cause of query delays.
 
 Combine network stats with wait statistics to diagnose if network problems are causing SQL Server waits:
 
@@ -152,9 +156,9 @@ WHERE w.wait_type LIKE '%NETWORK_IO%';
 
 ### F. Find interfaces with high collision counts
 
-Detect and address network segments with excessive collisions, which might degrade SQL Server performance.
+Detect and fix network segments with excessive collisions, which might slow down SQL Server performance.
 
-Collisions can indicate network congestion or misconfigured hardware:
+Collisions can show network congestion or misconfigured hardware:
 
 ```sql
 SELECT interface,
@@ -165,9 +169,9 @@ WHERE tx_collisions > 0;
 
 ### G. Baseline and alerting example
 
-Integrate with monitoring tools to proactively notify database administrators of potential network problems before they affect SQL Server workloads.
+Integrate with monitoring tools to proactively notify database administrators about potential network problems before they affect SQL Server workloads.
 
-Use this DMV in automated monitoring scripts to alert when error or drop counts increase unexpectedly.
+Use this DMV in automated monitoring scripts to alert when error or drop counts unexpectedly increase.
 
 Alert if any interface has more than 10 errors or drops:
 
@@ -203,4 +207,5 @@ WHERE recv_errors > 10
 
 - [sys.dm_os_linux_cpu_stats (Transact-SQL)](sys-dm-os-linux-cpu-stats-transact-sql.md)
 - [sys.dm_os_linux_disk_stats (Transact-SQL)](sys-dm-os-linux-disk-stats-transact-sql.md)
+- [sys.dm_os_linux_vm_stats (Transact-SQL)](sys-dm-os-linux-vm-stats-transact-sql.md)
 - [Performance best practices and configuration guidelines for SQL Server on Linux](../../linux/sql-server-linux-performance-best-practices.md)

docs/relational-databases/system-dynamic-management-views/sys-dm-os-linux-vm-stats-transact-sql.md

@@ -0,0 +1,107 @@
+---
+title: "sys.dm_os_linux_vm_stats (Transact-SQL)"
+description: sys.dm_os_linux_vm_stats returns a table with detailed Linux CPU statistics, offering system-level insights beyond SQL Server.
+author: rwestMSFT
+ms.author: randolphwest
+ms.date: 01/19/2026
+ms.service: sql
+ms.subservice: system-objects
+ms.topic: "reference"
+f1_keywords:
+  - "dm_os_linux_vm_stats"
+  - "sys.dm_os_linux_vm_stats_TSQL"
+  - "dm_os_linux_vm_stats_TSQL"
+  - "sys.dm_os_linux_vm_stats"
+helpviewer_keywords:
+  - "sys.dm_os_linux_vm_stats dynamic management view"
+dev_langs:
+  - "TSQL"
+---
+# sys.dm_os_linux_vm_stats (Transact-SQL)
+
+[!INCLUDE [sqlserver2025-linux](../../includes/applies-to-version/sqlserver2025-linux.md)]
+
+Returns Linux operating system-level virtual memory statistics, including metrics related to SQL Server and other processes running on the system, in [!INCLUDE [sssql25-md](../../includes/sssql25-md.md)] Cumulative Update (CU) 1 and later versions.
+
+| Column name | Data type | Nullable | Description |
+| --- | --- | --- | --- |
+| `vm_metric_name` | **nvarchar(256)** | No | Virtual memory metric name. |
+| `count` | **bigint** | No | Corresponding statistic for that metric. |
+
+## Permissions
+
+Requires `VIEW SERVER PERFORMANCE STATE` permission on the server.
+
+## Remarks
+
+`sys.dm_os_linux_vm_stats` provides system-wide memory observability to help you analyze memory pressure, page faults, reclaim activity, non-uniform memory access (NUMA) behavior, and to correlate SQL Server performance with overall OS memory health.
+
+Each row represents a single virtual memory metric, typically sourced from Linux interfaces such as `/proc/vmstat`. Metric availability and meaning can vary by Linux distribution, kernel version, and configuration.
+
+Use this DMV with other Linux-specific DMVs for holistic monitoring:
+
+- [sys.dm_os_linux_cpu_stats](sys-dm-os-linux-cpu-stats-transact-sql.md)
+- [sys.dm_os_linux_disk_stats](sys-dm-os-linux-disk-stats-transact-sql.md)
+- [sys.dm_os_linux_net_stats](sys-dm-os-linux-net-stats-transact-sql.md)
+
+### Usage scenarios
+
+Common scenarios for using `sys.dm_os_linux_vm_stats` include:
+
+- Investigating out-of-memory (OOM) events on Linux hosts.
+- Correlating SQL Server memory symptoms with OS reclaim activity.
+- Understanding NUMA-related memory behavior on multinode systems.
+- Performing deep OS-level observability directly from Transact-SQL.
+
+## Examples
+
+### A. View all virtual memory statistics
+
+The following query returns all available virtual memory metrics reported by the Linux kernel:
+
+```sql
+SELECT *
+FROM sys.dm_os_linux_vm_stats;
+```
+
+### B. Identify page fault activity
+
+The following query highlights page fault-related metrics, which can help identify memory pressure or inefficient memory access patterns:
+
+```sql
+SELECT vm_metric_name,
+       count
+FROM sys.dm_os_linux_vm_stats
+WHERE vm_metric_name IN ('pgfault', 'pgmajfault');
+```
+
+### C. Monitor locality of NUMA memory
+
+Returns NUMA-related virtual memory metrics to help understand memory locality across nodes:
+
+```sql
+SELECT vm_metric_name,
+       count
+FROM sys.dm_os_linux_vm_stats
+WHERE vm_metric_name LIKE 'numa%';
+```
+
+### D. Analyze activity for memory reclaim and compaction
+
+The following query helps you diagnose memory reclaim behavior and compaction pressure on the system:
+
+```sql
+SELECT vm_metric_name,
+       count
+FROM sys.dm_os_linux_vm_stats
+WHERE vm_metric_name LIKE 'pgsteal%'
+      OR vm_metric_name LIKE 'pgscan%'
+      OR vm_metric_name LIKE 'compact%';
+```
+
+## Related content
+
+- [sys.dm_os_linux_cpu_stats (Transact-SQL)](sys-dm-os-linux-cpu-stats-transact-sql.md)
+- [sys.dm_os_linux_disk_stats (Transact-SQL)](sys-dm-os-linux-disk-stats-transact-sql.md)
+- [sys.dm_os_linux_net_stats (Transact-SQL)](sys-dm-os-linux-net-stats-transact-sql.md)
+- [Performance best practices and configuration guidelines for SQL Server on Linux](../../linux/sql-server-linux-performance-best-practices.md)