make get_current_servable_instance async

[abrarsheikh]

This PR updates the injected dependency used in FastAPI routes to be asynchronous. Specifically, it modifies the implementation of get_current_servable_instance to be an async def so that FastAPI awaits it instead of offloading it to a threadpool. This change avoids unnecessary threadpool overhead and improves request performance.

Performance Comparison Table (With fix vs Master)

Metric	With fix	Master
# Requests	22,429	12,970
# Fails	0	0
Median (ms)	250	340
95%ile (ms)	310	430
99%ile (ms)	350	480
Average (ms)	248.04	338.4
Min (ms)	119	156
Max (ms)	392	662
Average size (bytes)	15	15
Current RPS	400.2	295.7
Current Failures/s	0	0

evidence that fastapi dependency injection is using threadpool

master

after fix

app used for benchmark

app = FastAPI()

@serve.deployment(max_ongoing_requests=10000)
@serve.ingress(app)
class MyDeployment:
    @app.get("/")
    async def request(self):
        return "Hello, world!"


app = MyDeployment.bind()

run using serve run app:app

device used: m6a.2xlarge
Concurrency: 100

[gemini-code-assist]

Summary of Changes

Hello @abrarsheikh, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

I've updated the get_current_servable_instance dependency injection in FastAPI routes to be asynchronous. This change prevents FastAPI from offloading this function to a threadpool, which significantly reduces overhead and improves request performance, as demonstrated by the provided performance metrics.

Highlights

• Asynchronous Dependency Injection: The get_current_servable_instance function, used as a dependency in FastAPI routes, has been converted to an async def function.
• Performance Improvement: This modification avoids unnecessary threadpool usage by FastAPI for this dependency, leading to improved request performance, higher requests per second (RPS), and lower latency (median, 95th, and 99th percentile).

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments or fill out our survey to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request correctly changes get_current_servable_instance to an async function. This is a solid performance optimization for FastAPI-based deployments, as it avoids the overhead of running the dependency in a threadpool. The change is well-explained in the description and backed by performance data. The implementation is simple, correct, and has no side effects. I approve this change.

[zcin]

🔥