The cloud has changed the business landscape—continuous innovation is no longer optional, and enterprise apps must modernize or perish. Generative AI is being adopted quickly not only to build net-new apps, but also to transform existing apps to deliver higher value AI-assisted outcomes. App modernization means updating and improving apps so they’re able to work well with evolving software and cloud. There are several ways to achieve this, whether by a simple ‘lift and shift’ of applications to the cloud or rewriting them as cloud-based services that work well with other modern solutions.

Today, we are excited to announce general availability of Sidecars in Azure App Service, a versatile pathway for organizations to modernize existing apps and add various kinds of powerful new capabilities without having to significantly rewrite the main application code. 

What are sidecars, and where could they fit into your modernization story?

What is sidecar pattern? Think of it like a motorcycle with an attached sidecar. You pair your primary enterprise app (the motorcycle) with a modern, supplementary set of tasks placed in their own container (the sidecar). These tasks enable you to add new capabilities by adding SDKs—like AI, logging, monitoring, and security features—to your primary application without the need to significantly modify and redeploy the app.

The sidecar pattern enables you to bring modern capabilities such as AI to your legacy apps much faster without rebuilding the primary app from the ground up. There are lifecycle and integration advantages as well: the primary app and sidecar capabilities can be treated and managed as a single unit. They’re easier to maintain since they remain distinct apps, and you can modify components of the sidecar independently without changing the entire app (and vice versa).

Using sidecars to add modern features to your existing apps

Let’s walk through a few app modernization scenarios where the sidecar approach may help.

Infuse generative AI into existing apps. Microsoft offers multiple AI models to help you get started in your AI journey, whether you choose a large language model (LLM) such as Azure OpenAI, or a small language model (SLM) like Phi-3. SLMs are lightweight, efficient, and accessible, making them ideal for limited computational resources and real-time inference. They can be fine-tuned for specific domains, enhancing performance and accuracy. SLMs are more secure, environmentally friendly, and suitable for edge computing due to their smaller codebases, lower energy use, and faster inference times. In this example, you can integrate Phi-3 ONNX using a sidecar pattern to deploy and run intelligent apps on Azure App Service without utilizing GPUs. This approach can offer a faster way to get started with AI with relatively lower cost. It also serves as an onramp to scale up to more generalized and powerful large language models (LLMs) such as deploying OpenAI-powered apps on Azure App Service in the future.

Gain modern observability, reporting, and analytics. Observability has become crucial for modern AI-infused applications. A sidecar gives you the ability to integrate functionality for metrics, logging, and tracing without having to completely rebuild the primary app. This includes first party solutions such as Azure Monitor Application Insights, a feature of Azure Monitor that excels in application performance monitoring (APM) for live web applications, as well as Azure Native ISV service partners such as Dynatrace, Datadog, and others. Learn more. 

Improve web app performance during high traffic. Redis Cache is an in-memory data store, which means it can retrieve data much faster than traditional databases. This can lead to application performance improvements, especially for read-heavy workloads. Azure Cache for Redis integrates seamlessly with other Azure services, such as Azure App Service, Azure Kubernetes Service (AKS), and Azure Functions. By deploying Redis cache as a sidecar, you can more easily unlock the benefits of caching and enable improved performance for your web applications.

These are just a few examples of how you can use sidecar to modernize your applications. The possibilities are only limited by your imagination!

Get started today with Azure App Service

Visit the product documentation to learn more and get started. Watch the learn live session on-demand for a demo of the new features from our experts. Join us at Microsoft Ignite 2024 and discover solutions that help modernize and manage intelligent apps, including demos of sidecar pattern and an opportunity to connect with the engineers building this technology in the Expert meet-up areas. 

If you’re new to Azure, take advantage of our free credit program by signing up for a free account at Azure App Service.

Here are some other helpful resources to learn more:

Catch A Glimpse into the Future: The Sidecar Pattern on Linux App Service
Visit Microsoft Learn to watch a tutorial on configuring a sidecar container
Read about building smarter apps by integrating Phi-3 SLM with Linux App Service
Learn how to implement local RAG using Phi-3 ONNX Runtime and Sidecar Pattern on Linux App Service
Revisit many of the links featured in this blog:

Optimizing SLM with ONNX Runtime: Phi-3 on CPU with Sidecars for App Service – Azure App Service
A Step-by-Step Guide to Datadog Integration with Linux App Service via Sidecars – Azure App Service
Powering Observability: Dynatrace Integration with Linux App Service via Sidecars – Azure App Service
Leveraging Redis as a Sidecar for Linux App Service – Azure App Service
Deploying OpenAI-powered apps on Azure App Service

​Microsoft Tech Community – Latest Blogs –Read More