News

.NET Core 2.1 Powers 34 Percent Bing Performance Boost

The modernization of the .NET ecosystem is paying off for Microsoft's Bing.com engineering team, which reports a 34 percent performance boost for the search engine following a shift from .NET Framework to .NET Core 2.1

By "eating its own dogfood," as the saying goes, Microsoft also reports runtime-adoption agility and start-up time gains for Bing.com.

Those all resulted from maturation of the .NET Core initiative, which started with .NET Core 1.0 in June 2016. It's a remake of the staid, 16-year-old, Windows-only .NET Framework, taking .NET cross-platform, modular and open source, with more flexible deployment options. It works hand-in-hand with .NET Standard, which defines all the APIs that should be available to all .NET implementations.

The Bing engineering team began moving to .NET Core/.NET Standard early, but was stymied by the lack of supported APIs in v1.x releases. The release of .NET Standard 2.0 added more than 20,000 APIs to the initiative, and the team began to gain traction in its migration off .NET Framework.

"Since its beginning, Bing.com has run on the .NET Framework, but it recently transitioned to running on .NET Core," explained Mukul Sabharwal of the Bing.com engineering team in a Monday (Aug. 20) blog post. "The main reasons driving Bing.com's adoption of .NET Core are performance (a.k.a serving latency), support for side-by-side and app-local installation independent of the machine-wide installation (or lack thereof) and ReadyToRun images.

"In anticipation of those improvements, we started an effort to make the code portable across .NET implementations, rather than relying on libraries only available on Windows and only with the .NET Framework. The team started the effort with .NET Standard 1.x, but the reduced API surface caused non-trivial complications for our code migrations. With the 20,000+ APIs that returned with .NET Standard 2.0, all that changed, and we were able to quickly shift gears from code modifications to testing. After squashing a few bugs, we were ready to deploy .NET Core to production."

By exploiting many .NET Core 2.1 performance improvements coming with its release in May, Sabharwal said, internal server latency over the last few months was improved by 34 percent:

Internal Server Latency Improvements
[Click on image for larger view.] Internal Server Latency Improvements (source: Microsoft).

He also cited gains with ReadyToRun images (code precompiled as a pre-deployment step rather than on local machines) and runtime agility, which resulted from "the ability to have an xcopy version of the runtime inside our application" that let the team adopt new runtime versions quicker.

The .NET Core 2.1 performance improvement that Sabharwal cited as being most impactful was better string manipulation, and he details several more in his post.

And, Sabharwal said, the team expects more to come: "We're excited about the future and are collaborating closely with the .NET team to help them qualify their future updates! The .NET Core team is excited because of our large catalog of functional tests and an additional large codebase to measure real-world performance improvements on, as well as our commitment to providing both Bing.com users fast results and our own developers working with the latest software and tools."

About the Author

David Ramel is an editor and writer at Converge 360.

comments powered by Disqus

Featured

  • Hands On: New VS Code Insiders Build Creates Web Page from Image in Seconds

    New Vision support with GitHub Copilot in the latest Visual Studio Code Insiders build takes a user-supplied mockup image and creates a web page from it in seconds, handling all the HTML and CSS.

  • Naive Bayes Regression Using C#

    Dr. James McCaffrey from Microsoft Research presents a complete end-to-end demonstration of the naive Bayes regression technique, where the goal is to predict a single numeric value. Compared to other machine learning regression techniques, naive Bayes regression is usually less accurate, but is simple, easy to implement and customize, works on both large and small datasets, is highly interpretable, and doesn't require tuning any hyperparameters.

  • VS Code Copilot Previews New GPT-4o AI Code Completion Model

    The 4o upgrade includes additional training on more than 275,000 high-quality public repositories in over 30 popular programming languages, said Microsoft-owned GitHub, which created the original "AI pair programmer" years ago.

  • Microsoft's Rust Embrace Continues with Azure SDK Beta

    "Rust's strong type system and ownership model help prevent common programming errors such as null pointer dereferencing and buffer overflows, leading to more secure and stable code."

  • Xcode IDE from Microsoft Archrival Apple Gets Copilot AI

    Just after expanding the reach of its Copilot AI coding assistant to the open-source Eclipse IDE, Microsoft showcased how it's going even further, providing details about a preview version for the Xcode IDE from archrival Apple.

Subscribe on YouTube

Upcoming Training Events