Azure AI Document Intelligence latency in metrics tab is showing less than 1 sec but in our application it's taking more than 5,6 sec even sometime more than 10 sec. is metrics tab data is wrong or any issue with my implementation?

Anurag Singh 0

Hi,

Azure AI Document Intelligence latency in metrics tab is showing less than 200ms but in our application it's taking more than 5,6 sec even sometime more than 10 sec for every request. is metrics tab data is wrong or any issue with my implementation?

I'm not able to understand exact problem. I'm using custom model of latest GA version.

Thanks

santoshkc 11,535 Reputation points Microsoft Vendor

2024-08-09T17:33:40.0433333+00:00

Hi @Anurag Singh,

Thank you for reaching out to Microsoft Q&A forum!

The discrepancy between Azure AI Document Intelligence's reported low latency and your application's higher latency is likely due to factors such as network conditions, data transfer time and potential variations in service performance. The metrics tab typically shows server-side processing time, while your application’s total latency includes additional delays.

To diagnose and address the issue, check network latency, data transfer time and ensure your application’s configuration is efficient. Review your service’s settings and consider any potential queuing or throttling effects.

I hope you understand! Thank you.
santoshkc 11,535 Reputation points Microsoft Vendor

2024-08-12T08:55:38.41+00:00

Hi @Anurag Singh,

We haven’t heard from you on the last response and was just checking back to see if the given response was helpful. Thank you.
romungi-MSFT 48,221 Reputation points Microsoft Employee

2024-08-21T07:53:13.74+00:00

@Anurag Singh Did you get a chance to check there any delays on client side on the application server while processing the requests? Thanks!!

Anurag Singh 0

HI @romungi-MSFT

the time in mentioned above was only api call time. By this way i'm calling your api.

DocumentAnalysisClient docAnalyzer = new DocumentAnalysisClientBuilder()
        .credential(new AzureKeyCredential(azureFormRecognizerKey))
        .endpoint(azureFormRecognizerUrl)
        .buildClient();
SyncPoller<OperationResult, AnalyzeResult> analyzeDocumentPoller = docAnalyzer.beginAnalyzeDocumentFromUrl(modelId, blobUrl);
AnalyzeResult analyzeResult = analyzeDocumentPoller.getFinalResult();

romungi-MSFT 48,221 Reputation points Microsoft Employee

2024-08-27T08:14:55.5966667+00:00

@Anurag Singh W.r.t the metrics, if you are looking at latency metrics it is actually the time to process a request to start the operation and send a response to acknowledge the same. So, in this case i think the low number is actually the time taken to process your request to start the analyze operation. The actual time of analyze operation in this case is dependent on many factors. Here is the official documentation on latency for DI analyze requests.

Latency is the amount of time it takes for an API server to handle and process an incoming request and deliver the outgoing response to the client. The time to analyze a document depends on the size (for example, number of pages) and associated content on each page. Document Intelligence is a multitenant service where latency for similar documents is comparable but not always identical. Occasional variability in latency and performance is inherent in any microservice-based, stateless, asynchronous service that processes images and large documents at scale. Although we're continuously scaling up the hardware and capacity and scaling capabilities, you might still have latency issues at runtime.

Since this is an async operation the time the analyze is taking is dependent on many factors like scaling at resource or region level and this might be taking more time until the poller indicates a result is available. Also, the service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support. You might have to rewrite your application logic to display the result only after a poller returns a result instead of waiting for the result as soon as the request is made to analyze. Thanks!!

Share via

Azure AI Document Intelligence latency in metrics tab is showing less than 1 sec but in our application it's taking more than 5,6 sec even sometime more than 10 sec. is metrics tab data is wrong or any issue with my implementation?

Your answer