This article is contributed. See the original author and article here.
Microsoft partners like Airplane Solutions deliver transact-capable offers, which allow you to purchase directly from Azure Marketplace. Learn about this offer below:
Airplane Pricing: This SaaS solution handles large volumes of transactions quickly and accurately while providing real-time fare quotes for airlines, trains, and shipping companies. With Air Pricing, you can reduce GDS cost and dependencies by 75 percent of the distribution cost for direct channels. You can also integrate this tool with other inventory systems such as Amadeus, Navitaire, or airline passenger service systems.
This article is contributed. See the original author and article here.
Our customer got the following error message performing a backup very large database to a blog storage – Error-1117 the request could not be performed because of an I/O device error.
In this situation, you may reach this error depending on the size of the blob storage, in this situation, please, run the backup in stripping backup format “Azure Blob Storage block blobs have 200GB size limitation. If your database is bigger than 200GB, you should use stripped backup by providing several URL locations where parts of the backup will be placed” Native database backup in Azure SQL Managed Instance | Microsoft Learn
This article is contributed. See the original author and article here.
What is ML.NET
Machine learning (ML) is everywhere. We use ML empowered applications every day: when choosing the next TV series to watch based on Netflix recommendations for example, or when asking Alexa to play our favorite song. Soon every application on every platform will incorporate some ML capabilities, empowering the application itself and making it smarter.
This is the key reason why ML has been added into the .NET ecosystem a few years back, by creating an open-source framework (ML.NET) which enables developers to train, build and ship custom ML models for a wide range of scenarios (sentiment analysis, forecasting, recommendations and more). Since then, the framework has evolved a lot, incorporating new features, with the preview release of the latest version (ML.NET 3.0) being announced a few weeks ago.
Now, you could argue that there’s tons of ML frameworks over there, so what’s the difference between ML.NET and the most common frameworks for data scientists (like scikit-learn, pytorch or tidyverse)?
ML.NET is not designed for data scientists only: people with different levels of ML understanding can empower their .NET solutions with some kind of AI. And this is immediately clear when thinking about the languages used for ML models development in this framework (C# and F#) which are the standard languages for .NET ecosystem, but not the languages traditionally used for data science and modeling – Python and R.
Another characteristic of this framework, which makes explicit the mission of democratizing the art of ML, is the availability of a graphical user interface as a Visual Studio component (called Model Builder) to build and deploy ML models in a few clicks.
Also, key objective of this framework is to make ML consumable, by providing tools to automatize not only the model training process but also some of the classical MLOps tasks, like model maintenance (retraining) and integration of the model in a client application.
Model builder
We mentioned that one of the key differentiators of this framework is Visual Studio Model Builder feature. And it is also the best place to start for ML beginners that wish to train a custom model on their scenario and data, and then consume it in their application.
After installing the Model Builder component, you can open its user interface (UI) from Visual Studio by right clicking on the project item of your solution in Solution Explorer and then selecting “Add -> Machine Learning”.
At this point, Model Builder provides you with a step-by-step procedure to import data, train and evaluate custom ML models for the scenario that best suits your application’s needs.
You can choose between different kinds of scenarios, including:
Classical ML use cases – like classification or regression: Model Builder uses Automated ML to train multiple models from your data in parallel and then pick the best one according to pre-defined evaluation metrics.
Deep learning use cases – like image or text classification: Model Builder uses your data to fine tune pre-trained deep learning models (e.g. NAS-BERT or Inception).
In any case, once training is completed, Model Builder generates the code that you can use to re-train the model (for example to tune hyperparameters or update the dataset) and consume it in your app.
ML.NET API & AutoML
The ML.NET API enables you to integrate machine learning into your new or existing .NET applications by installing the Microsoft.ML NuGet package. This option might be for you if you’re familiar with ML and looking for ways to best leverage ML in your application, with the familiarity of the .NET platform in C# or F#.
ML.NET is supported on:
.NET
.NET Core 2.0 and above
.NET Framework 4.6.1 and above.
You can also automate the process of applying machine learning, known as AutoML through the API. The typical ML workflow includes the iterative steps of preprocessing, training, and evaluation, repeating these steps until it reaches the desired results. With each iteration optimization techniques applied during the training and evaluation phases select the best algorithm and hyperparameters. If you’re a beginner to ML but want to use AutoML in your application, you can use the defaults provided in the API and let AutoML handle the rest. The API also enables experienced users to extend the defaults and customize the model. The AutoML API can be found within the Microsoft.ML.AutoML package on NuGet.
Model deployment
When your model is complete, there are various options for deploying it to the cloud with Azure. We’ll walk through how you can deploy your model to the cloud as a web API in Azure App Service, Azure Container Apps or Azure Functions.
Consuming a trained model begins with saving and exporting it. An exported model can be used with any .NET application or Azure Function through the ML.NET API in C# or F#. To use it, you’ll first add the exported model to the project, then use the API to import and load the model and finally add input data to make predictions. Refer to the documentation on options and tutorials on how to export and load your models with either the ML.NET Model Builder or ML.NET API.
In an ASP.NET application, whether it’s a web app like Razor Pages or an API, you’d use a controller class to access your trained model to use any input from a HTTP request or return predictions in a response, in addition to other controller tasks. To deploy your ASP.NET application to Azure, first build it locally in Visual Studio, then deploy it to App Service or Azure Container App through Visual Studio’s Azure development workload.
With Azure Functions you can run small pieces of code in a serverless environment in the cloud via triggers, which will run the functions code and define how it’s invoked. This is a cost-effective option for consuming a model across various applications. One option for integrating a model into a function is through an HTTP trigger, which invokes the function via an HTTP request, where your trained model can use input data from an HTTP request and send predictions in a response. With the Azure development workload in Visual Studio, you can create and test a local function with your trained model, and finally deploy your function to Azure.
How to get started?
ML.NET enables all .NET developers any level of data science experience can build ML models with the tools and platform they know. With a variety of scenarios to choose from, you load data to build a model with the Model Builder in Visual Studio or the ML.Net API. You can build ML pipelines that automate iterative steps within the workflow, creating the best version of your model. ML powered projects can be deployed to various platforms, including the cloud where a deployed ML.NET model can consume new data and return results, enabling end users to leverage ML within your application.
This article is contributed. See the original author and article here.
We know how important it can be to see a product, test drive all its capabilities, and explore different features before you are ready to buy. With the Dynamics 365 Customer Service trials, you can quickly and easily experience the product for free. There is no credit card required or software to install! Here is what you get:
A tenant with preloaded sample data that you can modify, plus the ability to create new data
The ability to visualize the power of out-of-box analytics based on the sample data
A voice and chat channel experience for both the customer and the agent that simulates real contact center processes with no configuration
You’ll learn how these intelligent business applications go beyond traditional customer relationship management (CRM) systems to help you adapt, innovate, and delight every customer. You can even make customizations to the trial to see how the product would look within your organization.
In the trial, you’ll discover how to:
Engage customers and provide omnichannel support
Empower agents with the right information at the right time for quick resolutions
Monitor and improve operations with insights for increased productivity
Sign up, launch the product, and then follow along with curated tours that show the breadth and depth of Dynamics 365 Customer Service capabilities without any manual effort required.
Dynamics 365 Customer Service trial landing page
How to Sign up
You can sign up for a trial with a work account (onmicrosoft) here or a personal (Gmail, Outlook, Yahoo) account here. To learn more, watch this short demo.
There are a few differences in the trial experience depending on whether you log in with a work account or a personal account. With a personal account, you will have access to the full trial except the ability to do the following:
Extend the trial beyond 30 days
Buy now and purchase a license directly from the trial environment
Customize the PVA bot (the default out-of-box bot will work)
Enable some features like case swarming
But no matter which email you use to sign up, you will get the out-of-box Power Virtual Agents (PVA), omnichannel, unified routing, and other core D365 Customer Service capabilities as part of the trial. You can see the full list of licenses that come with the trial here.
Once you’re logged in and on the home page, you can watch a short video showcasing the all-encompassing value of engaging customers with Dynamics 365 Customer Service, and then start the tours.
Engage customers and provide omnichannel support
Voice tour
You can now simulate the end-to-end experience of a customer calling your company and how it will be routed to an agent. You can see real-time sentiment analysis, real-time transcription, knowledge articles, and agent scripts already populated.
Agent view of a customer call
Chat tour
Like the voice tour, you can explore how customers will interact with your company through your company’s chat widget. Just click the Chat demo and see how the PVA bot greets customers and how they are routed to agents.
Agent view of a customer chat
Empower agents with the right information at the right time for quick resolutions
Explore cases by seeing what it’s like to take a case in the unified agent workspace. See incoming items for agents across a wide spectrum of channels. While issues can come from any supported channel, Customer Service workspace helps agents stay on top of their work with one unified agent desktop.
Agent view of active cases
Monitor and improve operations
Delve into the depth and breadth of our insight capabilities that help supervisors learn about their operations and provide meaningful, actionable data.
Supervisor view of case insights
With these features, you can test drive all that Dynamics 365 Customer Service has to offer in seconds. Embark on your journey to empower your agents and set up your organization for success. And if you like what you see, and you logged in with a work account (onmicrosoft.com email), you can buy a license today!
This article is contributed. See the original author and article here.
Software fails, hardware breaks and you can run into issues when trying new things. Whenever we encounter an issue on Azure, we quickly open our favorite search engine, weed through the results, and look for answers pertaining to that issue. But what if I tell you that we have a blade in the Azure Portal that can help you diagnose and resolve issues, find authoritative troubleshooting resources (tools, guided steps, and articles), and easily get additional help as needed.
The Diagnose and Solve problems blade empowers Azure customers to troubleshoot and solve Azure service issues via service/resource health insights, automated troubleshooters (insight diagnostics), curated troubleshooting guides (common solutions), and additional troubleshooting tools provided by service teams.
This blade is available for all the azure resources. In this blog post, we will look at SQL Server on Azure VM as an example.
The diagnose and Solve blade contains 2 sections.
Common Problems
Troubleshooting Tools
Common Problems
Each common problem has a title, a category, and a brief description, making it easier for customers to make the right selection. Customers can also search or use filters to look for a problem or a tool. One can also group the problem by category.
The following are the various categories that have been listed.
Administration-Management
Backup-Restore
Data Services
HADR
Performance
Security
Setup-Licensing
SQL Connectivity
Change
Health
The page also contains information about the resource health and any change details associated with the resource in the last 72 hours.
Once the customer identifies the problem they are having, clicking on “Troubleshoot” button, will open a new page where you can tell us more about the problem you are experiencing. Depending on the information provided and the problem scenario chosen, we get information on ways to fix the issue.
I have chosen Backup and Restore as the problem and clicking on the problem takes me to a new page asking me to tell more about the problem that I am experiencing. It also provides me with further options to choose that match my specific problem.
I have chosen “Database Backup or Restore using URL (Azure Storage)” as the issue that matches my problem. The page provides me with a list of common causes for the issues related to backup or restore using URL.
Assuming, the error msg that I am getting is Error 3201: Backup fails with – Operating system error 50, clicking on the headline, provides me with the steps to resolve the issue.
Troubleshooting Tools
Clicking on the Troubleshooting Tools tab for the SQL Server on Azure VM provides you with 4 options.
Performance Diagnostics
Troubleshoot network connectivity
Evaluate Best Practices for your SQL Server using SQL Assessments
Repair or Redeploy Iaas extension
Performance Diagnostics
PerfInsights is a self-help diagnostics tool that can be run through the portal or as a standalone tool that collects diagnostic data, produces analytic reporting, summarizes system information, and produces a system log output that can help troubleshoot virtual machine performance problems in Azure for various scenarios including SQL Server, Windows OS, Linux, Azure storage, and more.
From a SQL Server on Azure VM perspective, we can use PerfInsights captures to review SQL Server findings, review storage analysis against Azure storage thresholds, resource reporting, review the system summary information that can make it a lot easier to document your Azure SQL Virtual Machine environment.
If you are experiencing performance problems with virtual machines, before contacting support, run this tool. More information can be found here.
Troubleshoot network Connectivity
The connection troubleshoot feature of Network Watcher provides the capability to check a direct TCP connection from a virtual machine to a virtual machine (VM), fully qualified domain name (FQDN), URI, or IPv4 address. Network scenarios are complex, they are implemented using network security groups, firewalls, user-defined routes, and resources provided by Azure. Complex configurations make troubleshooting connectivity issues challenging. Network Watcher helps reduce the amount of time to find and detect connectivity issues. The results returned can provide insights into whether a connectivity issue is due to a platform or a user configuration issue.
SQL best practices assessment provides a mechanism to evaluate the configuration of your Azure SQL VM for best practices like indexes, deprecated features, trace flag usage, statistics, etc. Assessment results are uploaded to your Log Analytics workspace using Microsoft Monitoring Agent (MMA).
Repair SQL IaaS Agent extension if the status is “Failed”. This operation will also help install latest SQL IaaS Agent extension if it is not already installed.
There is also a ” Force repair” option. Checking that will enable the repair button even if the SQL IaaS Agent extension is in “Succeeded” provisioning state. This allows you to upgrade your SQL IaaS Agent extension which is already in “Succeeded” provisioning state to the latest version.
From an azure SQL perspective, we have similar scenarios curated for both Azure SQL Database and Managed instance. We will explore them in another blog post.
I would encourage you all to take advantage of the “Diagnostics and Solve” blade to quickly identify your issues and resolve them. If one cannot resolve the issue, you can always use Azure Support to further troubleshoot it.
Recent Comments