Monat: Juni 2022

MLOps Blog Series Part 1: The art of testing machine learning systems using MLOps

Testing is an important exercise in the life cycle of developing a machine learning system to ensure high-quality operations. We use tests to confirm that something functions as it should. Once tests are created, we can run them automatically whenever we make a change to our system and continue to improve them over time. It is a good practice to reward the implementation of tests and identify sources of mistakes as early as possible in the development cycle to prevent rising downstream expenses and lost time.

In this blog, we will look at testing machine learning systems from a Machine Learning Operations (MLOps) perspective and learn about good case practices and a testing framework that you can use to build robust, scalable, and secure machine learning systems. Before we delve into testing, let’s see what MLOps is and its value to developing machine learning systems.

Figure 1: MLOps = DevOps + Machine Learning.

Software development is interdisciplinary and is evolving to facilitate machine learning. MLOps is a process for fusing machine learning with software development by coupling machine learning and DevOps. MLOps aims to build, deploy, and maintain machine learning models in production reliably and efficiently. DevOps drives machine learning operations. Let’s look at how that works in practice. Below is an MLOps workflow illustration of how machine learning is enabled by DevOps to orchestrate robust, scalable, and secure machine learning solutions.

Figure 2: MLOps workflow.

The MLOps workflow is modular, flexible, and can be used to build proofs of concept or operationalize machine learning solutions in any business or industry. This workflow is segmented into three modules: Build, Deploy, and Monitor. Build is used to develop machine learning models using an machine learning pipeline. The Deploy module is used for deploying models in developer, test, and production environments. The Monitor module is used to monitor, analyze, and govern the machine learning system to achieve maximum business value. Tests are performed primarily in two modules: the Build and Deploy modules. In the Build module, data is ingested for training, the model is trained using ingested data, and then it is tested in the model testing step.

1. Model testing: In this step, we evaluate the performance of the trained model on a separated set of data points named test data (which was split and versioned in the data ingestion step). The inference of the trained model is evaluated according to selected metrics as per the use case. The output of this step is a report on the trained model's performance. In the Deploy module, we deploy the trained models to dev, test, and production environments, respectively. First, we start with application testing (done in dev and test environments).

2. Application testing: Before deploying an machine learning model to production, it is vital to test the robustness, scalability, and security of the model. Hence, we have the "application testing" phase, where we rigorously test all the trained models and the application in a production-like environment called a test, or staging, environment. In this phase, we may perform tests such as A/B tests, integration tests, user acceptance tests (UAT), shadow testing, or load testing.

Below is the framework for testing that reflects the hierarchy of needs for testing machine learning systems.

Figure 3: Hierarchy of needs for testing machine learning systems.

One way to think about machine learning systems is to consider Maslow's hierarchy of needs. Lower levels of a pyramid reflect “survival,” and the true human potential is unleashed only after basic survival and emotional needs are met. Likewise, tests that inspect robustness, scalability, and security ensure that the system not only performs at the basic level but reaches its true potential. One thing to note is that there are many additional forms of functional and nonfunctional testing, including smoke tests (rapid health checks) and performance tests (stress), but they may all be classified as system tests.

Over the next three posts, we’ll cover each of the three broad levels of testing, starting with robustness and then moving on to scalability and finally, security.

For further details and to learn about hands-on implementation, check out the Engineering MLOps book, or learn how to build and deploy a model in Microsoft Azure Machine Learning using MLOps in the Get Time to Value with MLOps Best Practices on-demand webinar.

Source for images: Engineering MLOps book
Quelle: Azure

Azure powers rapid deployment of private 4G and 5G networks

As the cloud continues to expand into a ubiquitous and highly distributed fabric, a new breed of application is emerging: Modern Connected Applications. We define these new offerings as network-intelligent applications at the edge, powered by 5G, and enabled by programmable interfaces that give developer access to network resources. Along with internet of things (IoT) and real-time AI, 5G is enabling this new app paradigm, unlocking new services and business models for enterprises, while accelerating their network and IT transformation.

At Mobile World Congress this year, Microsoft announced a significant step towards helping enterprises in this journey: Azure Private 5G Core, available as a part of the Azure private multi-access edge compute (MEC) solution. Azure Private 5G Core enables operators and system integrators (SIs) to provide a simple, scalable, and secure deployment of private 4G and 5G networks on small footprint infrastructure, at the enterprise edge.

This blog dives a little deeper into the fundamentals of the service and highlights some extensions that enterprises can leverage to gain more visibility and control over their private network. It also includes a use case of an early deployment of Azure Kubernetes Services (AKS) on an edge platform, leveraged by the Azure Private 5G Core to rapidly deploy such networks.

Building simple, scalable, and secure private networks

Azure Private 5G Core dramatically simplifies the deployment and operation of private networks. With just a few clicks, organizations can deploy a customized set of selectable 5G core functions, radio access network (RAN), and applications on a small edge-compute platform, at thousands of locations. Built-in automation delivers security patches, assures compliance, and performs audits and reporting. Enterprises benefit from a consistent management experience and improved service assurance experience, with all logs and metrics from cloud to edge available for viewing within Azure dashboards.

Enterprises need the highest level of security to connect their mission critical operations. Azure Private 5G Core makes this possible by natively integrating into a broad range of Azure capabilities. With Azure Arc, we provide seamless and secure connectivity from an on-premises edge platform into the Azure cloud. With Azure role-based access control (RBAC), administrators can author policies and define privileges that will allow an application to access all necessary resources. Likewise, users can be given appropriate access to manage all resources in a resource group, such as virtual machines, websites, and subnets. Our Zero Trust security frameworks are integrated from devices to the cloud to keep users and data secure. And our complete, “full-stack” solution (hardware, host and guest operating system, hypervisor, AKS, packet core, IoT Edge Runtime for applications, and more) meets standard Azure privacy and compliance benchmarks in the cloud and on the enterprise edge, meaning that data privacy requirements are adhered to in each geographic region.

Deploying private 5G networks in minutes

Microsoft partner Inventec is a leading design manufacturer of enterprise-class technology solutions like laptops, servers, and wireless communication products. The company has been quick to see the potential benefit in transforming its own world-class manufacturing sites into 5G smart factories to fully utilize the power of AI and IoT.

In a compelling example of rapid private 5G network deployment, Inventec recently installed our Azure private MEC solution in their Taiwan smart factory. It took only 56 minutes to fully deploy the Azure Private 5G Core and connect it to 5G access points that served multiple 5G endpoints—a significant reduction from the months that enterprises have come to expect. Azure Private 5G Core leverages Azure Arc and Azure Kubernetes Service on-prem to provide security and manageability for the entire core network stack. Figures 1 and 2 below show snapshots from the trial.

Figure 1: Screenshot of logs with time stamps showing start and completion of the core network deployment.

Figure 2: Screenshot from the trial showing one access point successfully connected to seven endpoints.

Inventec is developing applications for manufacturing use-cases that leverage private 5G networks and Microsoft’s Azure Private 5G Core. Examples of these high-value MEC use cases include Automatic Optical Inspection (AOI), facial recognition, and security surveillance systems.

Extending enterprise control and visibility from the 5G core

Through close integration with other elements of the Azure private MEC solution, our Azure Private 5G Core essentially acts as an enterprise “control point” for private wireless networks. Through comprehensive APIs, the Azure Private 5G Core can extend visibility into the performance of connected network elements, simplify the provisioning of subscriber identity modules (SIMs) for end devices, secure private wireless deployments, and offer 5G connectivity between cloud services (like IoT Hub) and associated on-premises devices.

Figure 3: Azure Private 5G Core is a central control point for private wireless networks.

Customers, developers, and partners are finding value today with a number of early integrations with both Azure and third-party services that include:

Plug and play RAN: Azure private MEC offers a choice of 4G or 5G Standalone radio access network (RAN) partners that integrate directly with the Azure Private 5G Core. By integrating RAN monitoring with the Azure Private 5G Core, RAN performance can be made visible through the Azure management portal. Our RAN partners are also onboarding their Element Management System (EMS) and Service Management and Orchestrator (SMO) products to Azure, simplifying the deployment processes and have a framework for closed-loop radio performance automation.
Azure Arc managed edge: The Azure Private 5G Core takes advantage of the security and reliability capabilities of Azure Arc-enabled Azure Kubernetes Service running on Azure Stack Edge Pro. These include policy definitions with Azure Policy for Kubernetes, simplified access to AKS clusters for High Availability with Cluster Connect and fine-grained identity and access management with Azure RBAC. 
Device and Profile Management: Azure Private 5G Core APIs integrate with SIM management services to securely provision the 5G devices with appropriate profiles. In addition, integration with Azure IoT Hub enables unified management of all connected IoT devices across an enterprise and provides a message hub for IoT telemetry data. 
Localized ISV MEC applications: Low-latency MEC applications benefit from running side-by-side with core network functions on the common (Azure private MEC) edge-compute platform. By integrating tightly with the Azure Private 5G Core using Azure Resource Manager APIs, third-party applications can configure network resources and devices. Applications offered by partners are available in, and deployable from the Azure Marketplace.

It’s easy to get started with Azure private MEC

As innovative use cases for private wireless networks continue to develop and industry 4.0 transformation accelerates, we welcome ISVs, platform partners, operators, and SIs to learn more about Azure private MEC.

Application ISVs interested in deploying their industry or horizontal solutions on Azure should begin by onboarding their applications to Azure Marketplace.
Platform partners, operators, and SIs interested in partnering with Microsoft to deploy or integrate with private MEC can get started by reaching out to the Azure private MEC Team.

Microsoft is committed to helping organizations innovate from the cloud, to the edge, and to space—offering the platform and ecosystem strong enough to support the vision and vast potential of 5G. As the cloud continues to expand and a new breed of modern connected apps at the edge emerges, the growth and transformation opportunities for enterprises will be profound. Learn more about how Microsoft is helping developers embrace 5G.
Quelle: Azure