March 2016 Summaries

Paul Stovell discusses the transition from RavenDB to SQL Server for Octopus 3.x, highlighting the decision to utilize SQL Server as a document store while maintaining some advantages of a document database model. The approach involved structuring tables to store complete aggregate roots in JSON format, alongside specific fields for frequent queries, and using customized ID generation inspired by RavenDB. Stovell outlines methods for modeling relationships, handling migrations with DbUp, and deliberately managing database operations from .NET code without relying on traditional ORM tools like Entity Framework. The strategy has yielded improvements in performance and user comfort, despite the unconventional use of SQL Server as a document store, and anticipates further enhancement with SQL Server 2016's JSON support. Stovell acknowledges that while this setup suits their needs, it may not be ideal for applications requiring frequent write operations or horizontal database scaling.

Octopus Deploy 3.4 introduces significant advancements with its Early Access Program (EAP), allowing users to engage with new features before the full release. The update focuses on two major aspects: multi-tenant deployments and enhanced elastic environment and transient machine support. The Alpha 1 version enables users to experiment with tenant management, linking tenants to projects, configuring environments, and applying new security permissions, all accessible via the Octopus API and client library. It also introduces the concept of machine policies to manage machine lifecycles, including handling machines that are offline or come online, with the ability to customize health checks and automate responses. Users are encouraged to participate by providing feedback, with a 45-day Enterprise trial license available, although downgrading to earlier versions is not supported.

Paul Stovell discusses the transformative impact of Windows Server 2016's Nano Server and containers on application deployment for developers. He highlights how the traditional deployment model—where applications are packaged into ZIP files and deployed as processes—will evolve with the introduction of Nano Server. This new server option is notably smaller and faster, offering developers the flexibility to deploy applications as ZIP files, Docker containers, or entire VM images, enhancing efficiency in continuous delivery pipelines. Stovell suggests that the future of server environments will increasingly rely on Nano Server and .NET Core due to their minimal resource requirements and rapid deployment capabilities, predicting that traditional full .NET Framework and server with GUI will become legacy technologies. He points out the benefits of isolation and security provided by Nano Server, though noting some deployment configuration challenges, and envisions a future where most greenfield applications and environments will favor this lean setup for scalability and efficiency.

ASP.NET Core 1 introduces significant changes that streamline the build and deployment pipelines for .NET developers by making publishing a standardized feature of the platform. This shift eliminates the need for tools like OctoPack, as developers can now use the dotnet CLI to publish applications across various platforms, generating a folder with all necessary files for deployment. This standardization simplifies the process, requiring only an additional step to package the application, typically by zipping the published folder before deploying it with Octopus. Furthermore, TeamCity offers a new plugin that facilitates these commands, and similar support is expected from other build servers, marking a transformative period for .NET application deployment.