thanks for your time. I am using Azure form recognizer of Azure cognitive service. It works well for most of my users. However, for Chinese users, Azure China has not released this product.
So Chinese users must request the Azure global endpoint and the issue comes :
the latency is pretty high, sometimes , it comes out as 5000ms. So is there anyway to reduce the latency? Or any workaround ?
Any suggestion is appreciated. Thanks !
As #GauravMantri recommended, you could host it in a region close to China.
If it still comes out as high latency, you can try this way that recommended by the Microsoft partner consultant：
Deploy an APIM service in Azure China.
Put your Azure global endpoint behind the APIM and expose APIs by APIM. Let your Chinese users access the Azure global endpoint via APIM could greatly reduce the network latency.
I have tried it before and it works for me.
There is an Azure Form Recognizer that runs in containers.
https://learn.microsoft.com/en-us/azure/cognitive-services/form-recognizer/form-recognizer-container-howto
It is marked as "retiring". But maybe you could run it in AKS in China until Azure China has released this product.
Related
Is there anyway we can get notified in case of Azure PaaS service is down? example Storage account or Load balancer etc?
Thank you rickvdbosch. Posting your suggestion as an answer to help other community members.
You can set Health alerts. "Azure Service Health notifies you about Azure service incidents and planned maintenance so you can take action to mitigate downtime. Configure customizable cloud alerts and use your personalized dashboard to analyze health issues, monitor the impact to your cloud resources, get guidance and support, and share details and updates."
can we add specific service in the service health?
In Service Health, choose the Potential impact tab to see the specific list of resources you own that might be impacted by the issue. You can download a CSV list of these resources to share with your team.
I am testing Azure TTS service and it is working well on my dev PC.
Now, I am test on a windows server in a secure data center.
However, it may not work because of company firewall system which blocks both inbound/outbound traffic.
So, I need complete API address list to open firewall.
I am using C# and Azure Cognitive Service Nuget package.
I initialize SDK using “SpeechConfig.FromSubscription(key, region)”.
I found that some related address in Azure API help page and Github sample as the followings for southeastasia;
https://southeastasia.api.cognitive.microsoft.com/sts/v1.0/issueToken
https://southeastasia.tts.speech.microsoft.com/cognitiveservices/v1
Could you please let me know whether it is right?
Best regards.
Yea, that is right !
The endpoint is usually of the below type :
https://<REGION_IDENTIFIER>.api.cognitive.microsoft.com/ #for token
https://<REGION IDENTIFIER>.tts.speech.microsoft.com #for other communication
Region Identifier is the region shortened.
The API reference for text to speech is detailed here.
Alternatively, if you are looking to allow all Azure cloud traffic. The below would be complete list. This would be the recommended approach (The below is Public Cloud, For US gov you could refer this).
*.aadcdn.microsoftonline-p.com
*.aka.ms
*.applicationinsights.io
*.azure.com
*.azure.net
*.azurefd.net
*.azure-api.net
*.azuredatalakestore.net
*.azureedge.net
*.loganalytics.io
*.microsoft.com
*.microsoftonline.com
*.microsoftonline-p.com
*.msauth.net
*.msftauth.net
*.trafficmanager.net
*.visualstudio.com
*.windows.net
*.windows-int.net
If one application have Azure serviceBus, EventHub in diff Azure Namesapces, web application and also other azure services (eg: cognitive services). can these be accessed with one URL by using Gateway or Load balancer or traffic manager or any other option ?
My problem is - if we have diff namesapces, we need to whitelist every time when there is new Namespaces and it could so too much of a work. so wondering if we can have one common DNS/URL that would make life easier.
Today, Service Bus and Event Hubs don't support any sort of network gateway. This is due to fact that namespace in the connection string used for authorization purpose at the service side.
To add a bit of context to Serkant's statement, support for this scenario is something that is on our roadmap, and hopefully in the near term. Unfortunately, I don't have a date to share currently. The work is being tracked [here] should you wish to keep an eye on it.
I'm following the next tutorial: https://learn.microsoft.com/en-us/azure/azure-functions/functions-twitter-email. I had some problems with connecting to Twitter in Chrome, switched to Edge and was able to make the connection.
However for the sentiment analysis, created earlier in the tutorial, I'm constantly getting, in both browsers, the messages
Please check your account info and/or permissions and try again.
I thought first that it would take a while before the service would kick in gear so I took the same Key1 and went to the API overview from where people can test the service after providing the key. And that worked.
So I'm wondering on how to continue or how to solve the connection problem as the analysis service key works out.
As the time of writing this answer only cognitive accounts deployed into West US region are supported with the logic apps connector. However we are aware that cognitive services recently rolled out a number of new regions for text analytics, and are currently deploying the fix to support other regions. By the end of this week any region should be supported. For now, the workaround is likely best approach (deploy a West US-based cognitive service account).
I had the same issue when trying to login with Chrome.
Edge resolved this case and my resources are deployed in North Europe.
I had the same issue
Please check your account info and/or permissions and try again+text analytics+logic app
when trying establish connection with Text Analytics Cognitive Service from Logic Apps.
I had deleted Text Analytics API and Recreated in Wets US Region and its working fine.
Hej!
We have just started using Windows Azure and are now in the phase of designing our infrastructure. A question that I haven't really found a stright answer for is weather there is a limit on how many endpoints I can have per subscriptions. Some research told me 25 and then I found another place saying 150. I haven't found anything on MS offical Azure site or blog.
Does anyone know? and have the limit been confirmed?
Thanks in advance,
Lucas
I think you're confusing subscription with deployment (a subscription is really a billing model for your Azure resources: compute, storage, bandwidth, etc. A deployment will have a collection of VMs (or web/worker roles) living behind a single xxx.cloudapp.net namespace. You'd then configure endpoints at a deployment level. For a Virtual Machine deployment, you'll only worry about external-facing (input) endpoints, since VMs can communicate internally across all ports. For web/worker Cloud Service deployments, you'll also have input endpoints.
Regard the number of endpoints per deployment: This number has grown over the years, and will continue to evolve. I'm not sure of the current limit, but... It's very simple to create an endpoint with PowerShell. With a simple for-loop, you should be able to create endpoints until an error is thrown.