Merge pull request #88277 from Minewiskan/aas-metrics-update

Aas metrics update

Author: v-albemi

articles/analysis-services/analysis-services-logging.md

@@ -5,7 +5,7 @@ author: minewiskan
 manager: kfile
 ms.service: azure-analysis-services
 ms.topic: conceptual
-ms.date: 02/14/2019
+ms.date: 09/12/2019
 ms.author: owend
 ms.reviewer: minewiskan
 
@@ -62,7 +62,7 @@ Selecting **Engine** logs all [xEvents](https://docs.microsoft.com/analysis-serv
 
 ### All metrics
 
-The Metrics category logs the same [Server metrics](analysis-services-monitor.md#server-metrics) displayed in Metrics.
+The Metrics category logs the same [Server metrics](analysis-services-monitor.md#server-metrics) to the AzureMetrics table. If you're using query [scale-out](analysis-services-scale-out.md) and need to separate metrics for each read replica, use the AzureDiagnostics table instead, where **OperationName** is equal to **LogMetric**.
 
 ## Setup diagnostics logging
 
@@ -156,27 +156,53 @@ To view your diagnostic data, in Log Analytics workspace, open **Logs**  from th
 
 In the query builder, expand **LogManagement** > **AzureDiagnostics**. AzureDiagnostics includes Engine and Service events. Notice a query is created on-the-fly. The EventClass\_s field contains xEvent names, which may look familiar if you've used xEvents for on-premises logging. Click **EventClass\_s** or one of the event names and Log Analytics workspace continues constructing a query. Be sure to save your queries to reuse later.
 
-### Example query
-This query calculates and returns CPU for each query end/refresh end event for a model database and server:
+### Example queries
+
+#### Example 1
+
+The following query returns durations for each query end/refresh end event for a model database and server. If scaled out, the results are broken out by replica because the replica number is included in ServerName_s. Grouping by RootActivityId_g reduces the row count retrieved from the Azure Diagnostics REST API and helps stay within the limits as described in [Log Analytics Rate limits](https://dev.loganalytics.io/documentation/Using-the-API/Limits).
 
 ```Kusto
-let window =  AzureDiagnostics
-   | where ResourceProvider == "MICROSOFT.ANALYSISSERVICES" and ServerName_s =~"MyServerName" and DatabaseName_s == "Adventure Works Localhost" ;
+let window = AzureDiagnostics
+   | where ResourceProvider == "MICROSOFT.ANALYSISSERVICES" and Resource =~ "MyServerName" and DatabaseName_s =~ "MyDatabaseName" ;
 window
 | where OperationName has "QueryEnd" or (OperationName has "CommandEnd" and EventSubclass_s == 38)
 | where extract(@"([^,]*)", 1,Duration_s, typeof(long)) > 0
 | extend DurationMs=extract(@"([^,]*)", 1,Duration_s, typeof(long))
-| extend Engine_CPUTime=extract(@"([^,]*)", 1,CPUTime_s, typeof(long))
-| project  StartTime_t,EndTime_t,ServerName_s,OperationName,RootActivityId_g ,TextData_s,DatabaseName_s,ApplicationName_s,Duration_s,EffectiveUsername_s,User_s,EventSubclass_s,DurationMs,Engine_CPUTime
-| join kind=leftouter (
-window
-    | where OperationName == "ProgressReportEnd" or (OperationName == "VertiPaqSEQueryEnd" and EventSubclass_s  != 10) or OperationName == "DiscoverEnd" or (OperationName has "CommandEnd" and EventSubclass_s != 38)
-    | summarize sum_Engine_CPUTime = sum(extract(@"([^,]*)", 1,CPUTime_s, typeof(long))) by RootActivityId_g
-    ) on RootActivityId_g
-| extend totalCPU = sum_Engine_CPUTime + Engine_CPUTime
+| project  StartTime_t,EndTime_t,ServerName_s,OperationName,RootActivityId_g,TextData_s,DatabaseName_s,ApplicationName_s,Duration_s,EffectiveUsername_s,User_s,EventSubclass_s,DurationMs
+| order by StartTime_t asc
+```
+
+#### Example 2
 
+The following query returns memory and QPU consumption for a server. If scaled out, the results are broken out by replica because the replica number is included in ServerName_s.
+
+```Kusto
+let window = AzureDiagnostics
+   | where ResourceProvider == "MICROSOFT.ANALYSISSERVICES" and Resource =~ "MyServerName";
+window
+| where OperationName == "LogMetric" 
+| where name_s == "memory_metric" or name_s == "qpu_metric"
+| project ServerName_s, TimeGenerated, name_s, value_s
+| summarize avg(todecimal(value_s)) by ServerName_s, name_s, bin(TimeGenerated, 1m)
+| order by TimeGenerated asc 
 ```
 
+#### Example 3
+
+The following query returns the Rows read/sec Analysis Services engine performance counters for a server.
+
+```Kusto
+let window =  AzureDiagnostics
+   | where ResourceProvider == "MICROSOFT.ANALYSISSERVICES" and Resource =~ "MyServerName";
+window
+| where OperationName == "LogMetric" 
+| where parse_json(tostring(parse_json(perfobject_s).counters))[0].name == "Rows read/sec" 
+| extend Value = tostring(parse_json(tostring(parse_json(perfobject_s).counters))[0].value) 
+| project ServerName_s, TimeGenerated, Value
+| summarize avg(todecimal(Value)) by ServerName_s, bin(TimeGenerated, 1m)
+| order by TimeGenerated asc 
+```
 
 There are hundreds of queries you can use. To learn more about queries, see [Get started with Azure Monitor log queries](../azure-monitor/log-query/get-started-queries.md).
 

articles/analysis-services/analysis-services-monitor.md

@@ -5,7 +5,7 @@ author: minewiskan
 manager: kfile
 ms.service: azure-analysis-services
 ms.topic: conceptual
-ms.date: 07/26/2019
+ms.date: 09/12/2019
 ms.author: owend
 ms.reviewer: minewiskan
 
@@ -50,6 +50,10 @@ Use this table to determine which metrics are best for your monitoring scenario.
 |MemoryLimitLow|Memory: Memory Limit Low|Bytes|Average|Low memory limit, from configuration file.|
 |MemoryLimitVertiPaq|Memory: Memory Limit VertiPaq|Bytes|Average|In-memory limit, from configuration file.|
 |MemoryUsage|Memory: Memory Usage|Bytes|Average|Memory usage of the server process as used in calculating cleaner memory price. Equal to counter Process\PrivateBytes plus the size of memory-mapped data, ignoring any memory, which was mapped or allocated by the in-memory analytics engine (VertiPaq) in excess of the engine Memory Limit.|
+|private_bytes_metric|Private Bytes |Bytes|Average|The total amount of memory the Analysis Services engine process and Mashup container processes have allocated, not including memory shared with other processes.|
+|virtual_bytes_metric|Virtual Bytes |Bytes|Average|The current size of the virtual address space that Analysis Services engine process and Mashup container processes are using.|
+|mashup_engine_private_bytes_metric|M Engine Private Bytes |Bytes|Average|The total amount of memory Mashup container processes have allocated, not including memory shared with other processes.|
+|mashup_engine_virtual_bytes_metric|M Engine Virtual Bytes |Bytes|Average|The current size of the virtual address space Mashup container processes are using.|
 |Quota|Memory: Quota|Bytes|Average|Current memory quota, in bytes. Memory quota is also known as a memory grant or memory reservation.|
 |QuotaBlocked|Memory: Quota Blocked|Count|Average|Current number of quota requests that are blocked until other memory quotas are freed.|
 |VertiPaqNonpaged|Memory: VertiPaq Nonpaged|Bytes|Average|Bytes of memory locked in the working set for use by the in-memory engine.|

articles/analysis-services/analysis-services-vnet-gateway.md

@@ -5,7 +5,7 @@ author: minewiskan
 manager: kfile
 ms.service: azure-analysis-services
 ms.topic: conceptual
-ms.date: 01/09/2019
+ms.date: 09/12/2019
 ms.author: owend
 ms.reviewer: minewiskan
 
@@ -16,10 +16,12 @@ This article describes the **AlwaysUseGateway** server property for use when dat
 
 ## Server access to VNet data sources
 
-If your data sources are accessed through a VNet, your Azure Analysis Services server must connect to those data sources as if they are on-premises, in your own environment. You can configure the **AlwaysUseGateway** server property to specify the server to access all datasource data through an [On-premises gateway](analysis-services-gateway.md). 
+If your data sources are accessed through a VNet, your Azure Analysis Services server must connect to those data sources as if they are on-premises, in your own environment. You can configure the **AlwaysUseGateway** server property to specify the server to access all data sources through an [On-premises gateway](analysis-services-gateway.md). 
+
+Azure SQL Database Managed Instance data sources run within Azure VNet with a private IP address. If public endpoint is enabled on the instance, a gateway is not required. If public endpoint is not enabled, an On-premises Data Gateway is required and the AlwaysUseGateway property must be set to true.
 
 > [!NOTE]
-> This property is effective only when an [On-premises data gateway](analysis-services-gateway.md) is installed and configured. The gateway can be on the VNet.
+> This property is effective only when an [On-premises Data Gateway](analysis-services-gateway.md) is installed and configured. The gateway can be on the VNet.
 
 ## Configure AlwaysUseGateway property