Insight into your platform can be of huge benefit to both operations, developers and the security team, but logging to much can result in high costs. How can you optimize logging and monitoring to pay for only what you need. There are several ways to save on logging and some patterns and anti-patterns to be aware of to not overspend on logging. Let’s look at that now.
Before we start, I would like to thank Thomas Thornton and Joe Carlyle who are once again hosting the Azure Spring Clean!
On the 31st of August 2024, Microsoft will deprecate the Log Analytics VM Agent. This change has been announced years ago, but still many are not prepared. How can you assess your environment and migrate to the Azure Monitor Agent before the agent will stop working. In this 2 part blog service, we will look at how you can find and migrate away from the Log Analytics Agent before the VM extension gets deprecated.
Here is a collection of snippets of code used in the “Cost Optimization in the wild! - Experiences from reducing costs” presentation. If you are missing any snippets of code used in a demo, or if you have any other question, feel free to message me on Twitter or LinkedIn.
Orphaned disks Old Snapshots App Gateways and Load Balancers Storage v1 App Service Plans Orphaned disks KQL to find disks resources | where type == "microsoft.
In Microsoft’s Well-Architected Framework there is a pillar for Cost Optimization. Some of the principals for optimizing cost are to continuously look for and clean up orphaned resources like disks and public IPs. In this blog I will show you some KQL queries that will help you find these resources and considerations you should take before cleaning them up. You can then use these queries in an Azure Workbook or dashboard to continuously review your environment.
Last blogpost I showed you an Azure Policy that checked for a SQL server firewall rule. The results would be a compliance view in Azure Policy. After the blogpost, me and Dennes Torres had a chat about the results you get in the compliance view. The problem: you don’t get the resource ID or resource name for the SQL server with the setting on. This isn’t really a problem if you have only one not-compliant resource, but if you have several, it becomes tedious.
