I’ll admit it, I shot myself in the foot, not following the documentation to the letter, but I maintain that when configuring a feature with a potentially huge cost impact, “turn it off and hope you manage to turn it on right afterwards”, is really shoddy API design. That to configure the adaptive sampling, I had to turn it off and on again - and that you can turn it on wrong! However, when updating this particular instance of the Telemetr圜onfiguration, nothing actually changed - as it was completely decoupled from the rest of the Application Insights infrastructure. I got, seemingly, a fully functional Telemetr圜onfiguration. So that’s what I did, and it seemed to work fine. You inject an IOptionsMonitor, and though that you get the configuration classes. Public void Configure ( IApplicationBuilder app, IHostingEnvironment env, Telemetr圜onfiguration configuration ) Ĭan you spot the difference? I certainly thought I was doing 1:1 what the documentation said for a long time.ĪSP.NET Core heavily uses the Options pattern, which is how to inject configuration into your classes. You can configure Application Insights with your custom processor like this example from the official docs: In the Application Insights C# SDK, you can write what’s called a Telemetry Processor, which can process, or discard telemetry before it’s sent to the server. In our case, we’re using Azure Event Hubs, and the Azure SDK for that is very chatty with it’s http requests, particularly when processing hundreds of messages every second.Īpplication Insights doesn’t make excluding data to send easy, but it does make it possible. With prices this expensive, you’ll probably only want to collect the data you’re actually interested in. And Application Insights can collect a lot of gigabytes. Application Insights charges a whopping 2.76$ per ingested Gigabyte. Humio charges approximately 0.9$ per gb of ingested data, and Datadog charges 0.1$. Some of their competitors charge much less. Application Insights is (primarily) priced per GB of data ingested - and it’s pretty expensive. Every log statement, outgoing and ingoing http call, every burp and sneeze. To do this, it C# collects a lot of data. It then allows you to view this data in numerous ways, some of them rather clever. ![]() It collects log, http requests, exceptions - you name it. Or the alternative title, “How Microsoft shot me in one foot, and then gave me a gun so I could shoot my other one” Custom Filtering - or the time Microsoft shot me in the footĪzure Application Insights is an Azure service to help you monitor your applications. ![]() I’m satisfied with the conclusion to this, even though I still maintain that the API choice is pretty questionable Update - the documentation error has since been fixed, and Microsoft has reached out to me for a full refund.Įvery engineer I’ve talked to from Microsoft has been nothing but wonderful, but the official support channels are still slow (we’re talking weeks). How Azure Application Insights cost our company 4k USD in a couple of weeks Blog about software development in ClimateTech
0 Comments
Leave a Reply. |