In
this article, we will learn how to get started with Azure Metrics
Advisor. Metrics Advisor is part of the Azure Cognitive
Services family and can analyze time-series data without you having to create a
machine learning algorithm. Metrics Advisor takes the data, analyses it,
surfaces incidence and anomalies, and lets you create alerts for them.
Let’s start by creating a Metrics Advisor resource.
First, we’ll
select a resource group. Next, we’ll select the region for the service. Now,
we’ll fill in a name. We also need to select a pricing tier. Then check the
box, and that’s it.
This is the
Azure Cognitive Service resource for the Metrics Advisor. Here in Quickstart,
we can click here to Open the workspace.
This is the
Metrics Advisor portal. Metrics Advisor works by ingesting data and analyzing
it.
The first
thing that we need to do in the Metrics Advisor portal is to add a data feed
that provides data. We’ll use the built-in Azure SQL Database example data. We
can do that here by clicking on “Start a tour”.
This takes
us through connecting to the sample data source and the connection string. Then
it inputs a sample query to retrieve a record of data, which we can now
configure to be ingested. This data schema already adheres to the Metrics
Advisor requirements. Once the data feed is submitted for onboarding, it can
take 10 minutes or so.
Once the
Metrics Advisor has analyzed the data and we can view the metrics in the
data.
We can
explore this metric to fine-tune the anomaly detection. Here, on the right are
the slices of data from this metric, we can fine-tune the detection sensitivity
or set data boundaries.
We can do
the same on specific slices to teach the algorithm what to look for.
Additionally, we can set up real-time alerts for anomalies, which can invoke
e-mails, WebHooks, and Azure DevOps.
When you have fine-tuned the process, Metrics Advisor can
detect anomalies in future incoming data and you can analyze them.
No comments:
Post a Comment