Merge pull request #7 from Microsoft/springstone

Springstone

Deploy/Tenant/RB-Ops-Hourly.ps1

@@ -1,4 +1,4 @@
-#Average CPU usage calculated over 10 minutes for the last 1 hours
+#Average CPU usage calculated over 10 minutes for the last 1 hour
 .\RB-ProcessLogs.ps1 `
 		-ReportName "perfavgcpu" `
 		-dynamicQuery "Type=Perf CounterName=""% Processor Time"" TimeGenerated>=NOW-1HOURS | measure avg(CounterValue) by Computer interval 30MINUTE"
@@ -6,7 +6,7 @@
 #All detected threats based on threat status rank
 .\RB-ProcessLogs.ps1 `
 		-ReportName "securitydetectedthreats" `
-		-dynamicQuery "Type=ProtectionStatus ThreatStatusRank > 199 ThreatStatusRank != 470 | measure max(ThreatStatusRank) as Rank by Computer | Top 5000"
+		-dynamicQuery "Type=ProtectionStatus ThreatStatusRank > 199 ThreatStatusRank != 470 | measure max(ThreatStatusRank) as Rank by Computer"
 
 #All Windows security login failures in the past 1 hour
 .\RB-ProcessLogs.ps1 `
@@ -16,4 +16,4 @@
 #All Linux Syslog errors in the past 1 hour
 .\RB-ProcessLogs.ps1 `
 		-ReportName "linuxsyslogerrors" `
-		-dynamicQuery "Type=Syslog SeverityLevel=error TimeGenerated>NOW-1HOUR"
+		-dynamicQuery "Type=Syslog SeverityLevel=error TimeGenerated>NOW-1HOURS"

Documentation/Cost-Estimate.md

@@ -0,0 +1,32 @@
+# zMonitor
+
+## Cost Estimates
+
+Cost estimates for the zMonitor solution really depend on what you are monitoring, and how often.
+
+All pricing estimates in this document are based on region "West Europe".
+
+### Tenant / Subscription
+
+For tenants or subscriptions with a small footprint, a couple of VMs, the free versions should be sufficient. Keep an eye on Log Analytics log usage, as you may need change to the paid version, depending on the amount of data you are collecting. If you're processing logs, you can quickly exceed the 500NB daily limit for the free tier of Log Analytics.
+
+| Component                     | Assumptions               | Cost (monthly)     |
+| ----------------------------- | ------------------------- | ------------------ |
+| Log Analytics                 | Free tier                 | $ 0.00             |
+| Azure Automation              | 500 minutes (Free)        | $ 0.00             |
+|                               |                           | **$ 0.00**         |
+
+### Service Provider / Central
+
+| Component                     | Assumptions               | Cost (monthly)     |
+| ----------------------------- | ------------------------- | ------------------ |
+| Azure Storage Account (BLOB)  | 10 GB stored              | $ 0.20             |
+| Stream Analytics              | 1 Unit                    | $ 89.28            |
+| Azure Cosmos DB               | 2 GB stored, 400 RUs      | $ 24.31            |
+| Azure Automation              | 500 minutes (Free)        | $ 0.00             |
+|                               |                           | **$ 113.78**       |
+
+* [Optional] Power BI - assume you have a license for PowerBI Desktop.
+
+
+

Documentation/Deploy-ServiceProvider.md

@@ -17,22 +17,32 @@ The overall process for tenant monitoring for the service provider is:
 
 1. Receive tenant OMS logs as CSV in storage account container
 1. Use Stream Analytics to move the CSV into Cosmos DB (formerly DocumentDB)
-1. Run cleanup process through Azure Automation at least daily (cleans up the CSV container)
+1. Run cleanup process through Azure Automation at least daily (cleans up the CSV container, and archives processed CSVs)
 1. Visualize. This solution provides a work in progress PowerBI sample for viewing data. Viusalization can be done through any mechanism familiar to you, including existing tools as long as they can query Cosmos DB. PowerBI is provided for convenience as a starting point.
 
 ## Deployment
 
-Below are the basic steps required to deploy the service provider component of the solution, provided as interim guidance while working on the ARM template.
+Below are the basic steps required to deploy the service provider component of the solution, provided as interim guidance while working on the ARM template (currently limited by stream analytics configuration).
 
 What's needed to setup the service provider components of zMontior.
 
 * Storage Account (BLOBs)
+
+  Storage for the CSV logs, Hot Locally Redundant (LRS) BLOB storage is sufficient. Cold may work but hasn't been tested.
   * Two containers
-    * Main logs container - the container where the logs get dropped from subscriptions/tenants.
-    * Archive lgos container
+    * Main logs container
+
+      The container where the logs get dropped from subscriptions/tenants.
+
+    * Archive logs container
+
+      Long term retention of CSV logs, useful for later processing. Not directly required by this solution.
+
 * Azure Autoamtion
+
+  Runs the CSV cleanup and archiving jobs.
   * Deploy runbook: [RB-Ops-CleanupDaily][1]
-    * Schedule to run at least once a day
+    * Schedule to run at least once a day, recommended to run every hour or two
     * Update storage details in RB-Ops-CleanupDaily:
     ```PowerShell
     $StorageAccountName = "<STORAGE ACCOUNT>"
@@ -41,11 +51,19 @@ What's needed to setup the service provider components of zMontior.
     $StorageAccountKey = "<STORAGE ACCOUNT KEY>"
     ```
 * Azure Cosmos DB
-    * Create database and collection
+
+  Where the log data gets stored in JSON format, and where we report from. When querying Cosmos DB, we'll need the connection details, including the URI and key (read-only is sufficient) - both available under the "Keys" property under "Settings" on the Cosmos DB blade.
+    * Create database and create a collection
     * Remember to set Time To Live (TTL) - recommended to set to 7 days (604800 seconds)
-    * Scale according to number of tenants, a starting scale on a single partition is 2000 RUs. 
+
+      This auto-deletes records in Cosmos DB older than what's specficied in the TTL setting. This keeps the collection size constrained  and query performance reasonable. Adjust this according to your specific requirements. Remember, the original data is archived in the BLOB archive container.
+
+    * Scale according to number of tenants and query performance
+
+    Start scale on a single partition with 400 RUs. Increase RUs as query performance is impacted. Data ingest should not be impacted at 400 RUs as we add data in short bursts. 
+
 * Stream Analytics
-  * Configure input : storage account main logs containers
+  * Configure input : storage account main logs container
   * Configure output : Cosmos DB collection
   * Define the query:
     ```SQL
@@ -56,8 +74,11 @@ What's needed to setup the service provider components of zMontior.
     FROM
         [StorageContainerCSVs]
     ```
+  * Start the stream job
+
+  ![Stream Analytics - Running](images/centralStreamAnalytics.png)
 * Visualize - PowerBI
-  * Configure connection to CosmosDB
+  * Configure connection to CosmosDB using URI and key (read-only)
 
     NOTE: Use the datasource connector "DocumentDB (Beta)"
 

Documentation/Log-Reports.md

@@ -6,15 +6,15 @@ The queries included in this sample solution are designed to be flexible and hig
 
 ### Daily Reports
 
-* Alerts generated in the past 24 hours that are still open
-* All Windows VMs that require updates
-* All VMs that are missing critical updates
-* All VMs with more than 2GB RAM available on average
-* All computers with their most recent data
+* Alerts generated in the past 24 hours that are still open ("activealertscritical")
+* All Windows VMs that require updates ("anyupdatesrequired")
+* All VMs that are missing critical updates ("criticalupdatesrequired")
+* All VMs with more than 2GB RAM available on average ("vmswithover2gbramavailable")
+* All computers with their most recent data ("allvmsmostrecentdata")
 
 ### Hourly Reports
 
-* Average CPU usage calculated over 10 minutes for the last 1 hours
-* All detected threats based on threat status rank
-* All Windows security login failures in the past 1 hours
-* All Linux Syslog errors in the past 1 hour
+* Average CPU usage calculated over 10 minutes for the last 1 hours ("perfavgcpu")
+* All detected threats based on threat status rank ("securitydetectedthreats")
+* All Windows security login failures in the past 1 hours ("acctloginfailurepasthour")
+* All Linux Syslog errors in the past 1 hour ("linuxsyslogerrors")

README.md

@@ -1,11 +1,17 @@
 # zMonitor
 
+An Azure platform native monitoring solution that enables Azure monitoring across multiple tenants or subscriptions.
+
 ## Overview
 
-An Azure platform native monitoring solution that enables monitoring across multiple tenants or subscriptions.
+Problem statement: A service provider with 50 tenants with Azure subscriptions provisioned through CSP (Cloud Solution Provider), needing to consolidate operational telemetry to optimize running costs as well as deliver higher SLAs with a minimum amount of administrative overhead.
+
+Enter zMonitor, a platform for reporting based on Log Analytics data collected, quickly gaining insights across tenants or subscriptions. Gain insights on disks capacity status, VM performance - over or under utilized CPU/Memory/Disk/etc, security vulnerabilities - failed logons, update/patch status, application errors, etc.
 
 The primary goal is to utilize Azure native components, and deliver an as simple as possible, highly configurable and scalable, cost-effective monitoring solution. The driving force behind this solution was the need to be able to monitor Azure resources across tenants (for service providers), using Azure native tools. While there are many metrics and logs generated in Azure, surfacing this information across subscriptions/tenants and effectively monitoring proved challenging.
 
+For an indication of potential costs for the solution, review the [cost estimate](Documentation/Cost-Estimate.md) documentation.
+
 ## Solution
 
 For monitoring within subscriptions, OMS Log Analytics is leveraged as the native log and metric aggregation toolset in Azure. Using the free tier of Log Analytics will be sufficient for most cases, but depends on the number of resources being monitored and the metrics being collected.