Skip to content

Latest commit

 

History

History
59 lines (40 loc) · 2.57 KB

README.md

File metadata and controls

59 lines (40 loc) · 2.57 KB

Text Analytics for Health Container Async Batch Sample

This sample provides code examples and best practices on how to use the containerized version of Text Analytics for Health in a scalable way.

After completing the setup steps for this sample you will have:

  • A .Net Core console application to help you stress test the service.
  • A Kubernetes cluster that with several Text Analytics for Health Containers.
  • Azure Storage service with containers, Queues and Azure Table Storage.

All these elements can be setup through a guided tutorial with several deployment scripts. With the goal to be used in an scalable and asynchronous way.

Throughput and hardware recommendation

Before setting up the sample, it is important to understand the difference between the hosted and containerized service. Text Analytics for Health has a hosted and containerized version. There is a difference on throughput and limits for these deployment options

Throughput recommendations

Hosted Container
Max Documents Per Request 25 1000
Max characters Per Request 125,000 125,000
Document Throughput per second (S Tier) // 1000 characters per second

Harware recommendations

In the containerized version we recommend to use the following hardware specifications

Minimal Recommended
CPU 4 6
Memory 10 12

**IMPORTANT: The setup can currently only process documents where the processing does not take longer then 24 hours. The service will clear the Queue after 24 hours, this will be fixed in the upcoming version **

Architecture

The .Net Core Client Console Application sends one or more documents to the Azure Kubernetes Cluster (max 25). All documents and processed results are asynchronous processed and stored on an Azure Storage Account. If you are looking to process many documents, we recommend looking at the Azure Kubernetes Autoscaler to dynamically increase/decrease nodes based on requests "A diagram of the Intelligent dashboard architecture"

Prerequisites

  • An Azure Subscription
  • kubectl
  • az CLI installed

Setup the sample

The setup wil help you provision:

  • Text Analytics for Health
  • Azure Storage Account
  • Azure Kubernetes Service

Click here to setup the resources.

Using the sample

You can find all the relevant information on how to use the client sample for the containerized setup here