Analytics is a powerful search tool that lets you analyze large volumes of any JSON or CSV data. You can run a wide range of queries, including statistical and machine learning algorithms. You might be familiar with it as part of Application Insights, but you can also apply it to any stream of NoSQL data.
For example, let’s suppose you receive a data feed about flights. You could automate a daily analysis of route popularity and congestion. Analytics can run complex queries, including joins, aggregations, and statistical functions, to extract the necessary results. You can view the results in the range of charts available in Analytics. Or you could have Power BI run the queries each day, plot the results on maps, and present them on a website.
Analytics was originally designed as the powerful analysis tool for web application telemetry from Application Insights. But in this blog, we’ll focus on applying it to a separate stream of data.
Set up
To analyze your data with Analytics, you need an account in Microsoft Azure.
Sign in to the portal and set up a Storage resource in Azure. This is where you will put your data before it’s sucked into Analytics.
Create an Application Insights resource. Then navigate from there to the Analytics page.
Define your data source
Before you analyze some data, you need to tell Analytics about its format. Like we said, we’ll skip the Application Insights material for now, and go straight to defining our own data source.
This opens a wizard where you name the data source and define its schema. You can do that either by providing an explicit schema, or by uploading a small sample of my data – that’s usually easier.
In the flight data example, the files are in CSV format. The sample data file includes headers, and the schema is automatically inferred from it. You get the opportunity to update the inferred data types and field names if necessary.
Ingest data
Once you’ve defined a schema, you can upload data files as often as you like. Data files of hundreds of MB are easily handled by Analytics.
To ingest the data, it’s easiest to automate the process with a short script. The script uploads the data to Azure storage, and then notifies Analytics to ingest it. There’s a sample in the import documentation.
Run queries
Here’s a query to look for the top 10 destination airports by airlines.
| summarize count() by Destination_airport
| top 10 by count_ desc
| render piechart
Result:
The query language is powerful but easy to learn, and has a piped model in which each operator performs one task – much easier to work with than the nested SELECTs of SQL.
Join multiple tables
Analytics can ingest multiple sources and your queries can run joins over them.
For example, but wouldn’t it much nicer to see the airports identified by full name instead of by their codes? Let’s add a new data source, airportsData, which maps each airport code to its name and other information. Now we can perform a join on the tables:
airlineRoutes
| summarize routeCount = count()
by airportCode = destination_airport // rename field for join
| top 10 by routeCount count_ desc
| join kind=inner ( airportsData ) on airportCode
| project routeCount, airportCode // the fields we want
| render piechart
Augmenting Application Insights telemetry
The main job of Analytics is as the powerful query tool of Application Insights, which monitors the health and usage of your web applications. One of the reasons for importing data into Analytics is to augment the telemetry data. For example, to make the telemetry reports more readable, query URLs can be translated to page names.
Get started today
Analytics can be applied to your data today. Read detailed how-to here.
Whether you want to enrich your data or to analyze the logging data of your application, you can easily add a new data source and start ingesting the data. With a high-volume ingestion, you can now apply the power of Analytics query language to your own custom data.
As always, feel free to send us your questions or feedback by using one of the following channels:
· Try Application Analytics
· Suggest ideas and vote in Application Insights ideas
· Join the conversation at the Application Insights Community
Quelle: Azure
Published by