# Session Proposal - (KubeCon + CloudNativeCon North America 2022)

https://linuxfoundation.smapply.io/prog/kccncna2022/

**Proposal Submission Deadline**: Jun 3 2022 11:59 PM (PDT)

### Session Title

> Please use Title Case when inputting your title here. Please note, if you are a CNCF Graduated or Incubating project maintainer or a TAG, Working Group, or User Group lead, you will be sent a separate form to fill out to register your talk for the Maintainer Track. Please do not submit your Maintainer Talk into this CFP form.

Native Image Acceleration Experience with Harbor, Dragonfly and Nydus

### Session Description

> Please provide a detailed, focused description of what this session will cover, written in the third person. This is what will be used in the online schedule if accepted. Note: We will not select a submission that has already been presented at a previous CNCF or Linux Foundation event within the last year. If your submission is very similar to a previous talk, please include information on how this version will be different. 1000 character max.

Container image is already an important area of container infrastructure. Image pulling occupies most of the container startup lifecycle, and in large-scale container scaling scenarios, the network bandwidth occupied by image pulling and the single point of pressure on the image registry has become a non-negligible problem. Harbor provides an image storage solution in terms of security, identity and management，the P2P-based file distribution system Dragonfly can help reduce network latency when pulling image, and the Nydus accelerated image service features image data lazy fetching, security enhancements and many other improvements. This session presents how these services can be deeply integrated to provide a large-scale efficient, secure joint image solution with a smooth user experience for the container ecosystem.

### Benefits to the Ecosystem

> This is your chance to elaborate. Tell us how the content of your presentation will help better the ecosystem or anything you wish to share with the co-chairs and program committee. We realize that this can be a difficult question to answer, but as with the description, the relevance of your presentation is just as important as the content. Max of 1,000 characters.

Deep integration enables Harbor to have a complete set of capabilities for image cache warming and even faster image building, and the P2P distribution of image data and image acceleration service makes it possible to build a large-scale, efficient and secure image infrastructure. The session also presents best practices on how seamless integration between open source components can enhance the container image experience of cloud-native users. The design of a more open accelerated image build system allows rapid integration between different image acceleration solution providers, providing a reference for where the current image spec can be improved and better driving the evolution of next-generation container image.

------ DISCARDED DRAFT ------

Container image is one of the cloud-native infrastructures that has spawned the entire image lifecycle ecology from the build, storage, and distribution to runtime, and while the image design itself has not improved much since its inception. This session introduces a series of improvements in build, container startup performance, and other aspects of the Nydus acceleration image, and how the Harbor distribution service natively supports the acceleration image format from conversion to present.

The drawback with the current OCIv1 container image design is that the container must wait until the entire image is downloaded and decompressed before it can be started, and is also in a tight spot in terms of data de-duplication, global data storage optimization, and ensuring data consistency at runtime. Harbor as an image distribution service, deeply supports the smooth transition of acceleration image formats such as Nydus in the absence of the OCIv2 standard and provides the traceability of acceleration images at a native granularity. This session describes how Nydus and Harbor are providing users with an end-to-end next-generation image infrastructure experience.

--- By WangYan ---

With the rapid rise of cloud-native applications represented by containers and Kubernetes, more and more customers have accepted and recognized the concept of modern applications and migrated their business from traditional service to modern application platforms built on Kubernetes. For modern applications, container applications are an extremely critical core carrier. Poor image distribution efficiency and experience will not only directly affect the operation and maintenance process of the platform and the applications actually running, but will also bring a bad experience to platform operators and customers.Accelerating the large-scale and efficient distribution of images through innovative solutions is an inevitable requirement to achieve a better modern application operating platform, and a higher requirement to bring a better user experience for platform operators and customers. In this talk, we propose a full lifecycle OCI artifact(image) acceleration mechanism by leveraging the benefits offered by Harbor, such as pluggable accelerator, Peer-2-Peer(P2P) preheat to serve the Kubernetes cluster, and one of our goals is to improve the deployment performance in the Kubernetes cluster, especially for the situation of hundreds or thousands of pods.

Since business increasingly needs to respond quickly to real-world situations, containerized services are becoming popular, but unfortunately, deployment methods are very slow. The approach that we proposed can precisely address this pain point, and according to our observations, can highlight it much more efficiently. The audience will:

- Understand that the pluggable accelerator service, like Nydus, eStargz and accelerated artifacts.
- Know how harbor integrates with the accelerator service and describe the linkage of an OCI artifact and its accelerated artifacts to provide rich functionality of an artifact.
- Understand the artifact acceleration service specification, that defines an API protocol to facilitate and standardize the artifact acceleration service that converts OCI artifacts to accelerated images.
-