Understanding Azure ADX, Log Analytics, and Application Insights: Use Cases, Data Retention, and Benefits

Azure ADX (Kusto), Log Analytics (LA) & Application Insights (AI). These three offerings from azure are bit confusing to me. What is the situation when I will be using one over other. I understand when I provision an AppInsights underneath a LA is provisioned. I can query the data from both LA & AI. However, If I provision a new LA and point the AppInsights to it. I still see the data which was ingested before this change.
1. Does AppInsights has its own data store? then why does it need an LA.
2. If AppInsights does not have its own data store how I am seeing the data ingested in the previous LA.

I have around 10+ of AppInsights instances, is it recommended to point all of these to a single LA across subscriptions. Are there be benefit with this? Like stitching telemetry across AppInsights etc.?

When I switch over these AppInsights to a new LA, do I need to worry for the past data as I see the AppInsights still retains old logs even after switching to another AppInsights.

ADX seems to have its own data store, what are the scenarios where I need to provision this additional resource on top of LA. I am wondering bringing LA data to ADX will have any specific benefits in terms of application access or stitching telemetry.

That is a lot of questions you're asking! I'll try to get them answered.

I understand when I provision an AppInsights underneath a LA is provisioned.

Correct, that is the new way of doing things, unless you use a Classic Application Insights Resource.

However, If I provision a new LA and point the AppInsights to it. I still see the data which was ingested before this change.

The older Classic Application Insights Resource does not use a Log Analytics Workspace. You can convert such a resource to use a Log Analytics Workspace.
The data ingested before this moment is indeed still available but is subject to the retention settings as stated in the docs:

> Your classic resource data persists and is subject to the retention settings on your classic Application Insights resource.

That means that there will be a moment the old data isn't available anymore and all data is stored in the Log Analytics Workspace.

1. Does AppInsights has its own data store? then why does it need an LA. 2. If AppInsights does not have its own data store how I am seeing the data ingested in the previous LA.

The classic Application Insights resources did have their own storage and as such, did not need a Log Analytics Workspace. New workspace based resources do not have their own storage and rely on a Log Analytics Workspace.

Converted Application Insights resources do you Log Analytics Workspace but for a smooth transition the data before the conversion is still accessible until the retention period is elapsed.

I am wondering bringing LA data to ADX will have any specific benefits in terms of application access or stitching telemetry.

No, you can have multiple Application Insights resources pointing to the same Log Analytic workspace to make cross resource queries easier. But even before that you could already create such queries.

I have around 10+ of AppInsights instances, is it recommended to point all of these to a single LA across subscriptions. Are there be benefit with this? Like stitching telemetry across AppInsights etc.?

That is totally up to you. If you need to assign different permissions to different users to access data per application you might want to seperate the resources, otherwise it could be fine to use one Log Analytics workspace. As cross resource queries are already easy to create that is not a real benefit I think.

Azure ADX (Kusto), Log Analytics (LA) & Application Insights (AI). These three offerings from azure are bit confusing to me. What is the situation when I will be using one over other.

Application Insights depends on Log Analytics and is part of Azure Monitor. It is mend to be used for application diagnostics, logging and traces amongst others.

Azure ADX is an offering geared towards real-time analysis on large volumes of data streaming from applications, websites, IoT devices, and more. Its purpose is analyzing business data.