[KV offload][1/N] Introduce an offloading component

[orozery]

This PR adds a new offloading component, composed of:

1. A scheduler side OffloadingManager (abstract) which kicks-off KV data transfers and keeps track of offloaded data.
2. A worker side OffloadingQueueManager which asynchronously manages KV transfers.

Part of the work described in RFC #19854

[gemini-code-assist]

Summary of Changes

Hello @orozery, 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!

This pull request introduces a new, dedicated offloading component for vLLM v1, establishing the core infrastructure for managing and transferring Key-Value (KV) data. It sets up abstract interfaces for the scheduler to oversee offloaded data and provides concrete mechanisms for workers to perform these transfers asynchronously, laying the groundwork for more flexible memory management.

Highlights

• New Offloading Component: Introduces a foundational offloading module designed to manage and facilitate asynchronous Key-Value (KV) data transfers between different storage mediums within vLLM v1.
• Scheduler-side OffloadingManager: Defines an abstract OffloadingManager class intended for the scheduler, responsible for tracking offloaded data, preparing load/store operations, and managing the lifecycle of KV blocks, including lookup, touch, and eviction protection.
• Worker-side OffloadingQueueManager: Implements the OffloadingQueueManager on the worker side to asynchronously manage KV data transfers. This manager utilizes dedicated OffloadingWorker threads to handle different types of transfers concurrently.
• Load/Store Specifications: Introduces LoadStoreSpec as an abstract base class for defining metadata necessary for KV data transfers. Concrete implementations like GPULoadStoreSpec and CPULoadStoreSpec are provided as examples for GPU and CPU memory locations.
• Test Coverage: Adds a new test suite (tests/v1/offloading/test_worker.py) to thoroughly validate the functionality of the OffloadingQueueManager and its asynchronous transfer capabilities, ensuring reliability of the new component.

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 is currently in preview and 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 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 PR introduces a new offloading component with an abstract OffloadingManager and a concrete OffloadingQueueManager. Key feedback includes updating a test docstring, improving error handling in OffloadingWorker, ensuring thread-safety for job ID generation, and considering robust resource management for OffloadingQueueManager.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[ApostaC]

Thanks for the effort! I really enjoy the overall interface design!
Left a few questions & comments for some design details.

[ApostaC]

Thanks for all the clarification! LGTM!

[njhill]

Thanks @orozery

[njhill]

Suggested change

	parepare_load() - prepare given blocks to be read.
	This given blocks will be protected from eviction.
	prepare_load() - prepare given blocks to be read.
	The given blocks will be protected from eviction.

[pytorch-bot]

No ciflow labels are configured for this repo.
For information on how to enable CIFlow bot see this wiki