NVM-Based Container Image Layering for Deployment Acceleration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The slow deployment and startup of containers in traditional Docker architectures due to disk I/O limitations, which is exacerbated by the high latency and I/O burst during mirror image downloading and storage, necessitates an optimization method that leverages Non-Volatile Memories (NVMs) to enhance performance while minimizing write wear and usage.
Innovation Solution
The method involves classifying each image layer as either Layer Above Last Download Layer (LAL) or Layer Below Last Download Layer (LBL), storing LALs in non-volatile memory and selectively storing LBLs in either non-volatile memory or hard drive, acquiring and storing hot image files in non-volatile memory, and sorting mirror images by access frequency to optimize storage and reduce write wear.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If mirror images are stored in hard drive, then storage capacity is sufficient, but deployment and startup speed is slow due to disk I/O limitations
Solution Approach 1:
The patent segments mirror images into multiple image layers and stores them in a distributed manner across NVM and hard drive. By dividing the monolithic mirror image storage into layered segments, the system can selectively store frequently accessed layers in high-speed NVM while keeping less frequently accessed layers on hard drive, thereby reducing overall I/O wait time and accelerating container deployment and startup.
Solution Approach 2:
The patent introduces NVM as an intermediary storage layer between the hard drive and the container runtime system. This intermediary buffer storage accelerates access to mirror image layers during container deployment and startup, reducing the direct I/O burden on the hard drive and eliminating the bottleneck that slows down container operations.
2Speed
If NVM is used for storing mirror image layers, then deployment and startup speed is improved, but write wear on NVM increases
Solution Approach 1:
The patent applies local quality by differentiating storage locations based on access patterns of different image layers. Frequently accessed layers (those containing hot image files) are stored in NVM for fast access, while less frequently accessed layers are stored on hard drive. This localized optimization ensures that NVM write operations are minimized and concentrated only on necessary layers, preserving NVM write endurance while maintaining deployment speed.
Solution Approach 2:
The patent implements partial action by selectively storing only critical image layers (LALs and hot image files) in NVM rather than all layers. This partial utilization of NVM capacity reduces the total write volume on NVM, thereby extending its write endurance while still achieving the performance benefit of accelerated container operations for the most frequently accessed layers.
3Speed
If all image layers are stored in NVM, then access speed is maximized, but storage space consumption increases
Solution Approach 1:
The patent implements partial action by storing only essential image layers (LALs and hot image files) in NVM rather than all layers. This selective partial storage achieves the performance benefit of fast access for critical layers while consuming minimal NVM capacity, avoiding the excessive space consumption that would result from storing all layers in NVM.
Solution Approach 2:
The patent applies local quality by optimizing storage location based on the specific access characteristics of different image layers. Critical layers that require fast access are placed in NVM, while non-critical layers remain on hard drive. This localized quality differentiation maximizes the utility of limited NVM space while maintaining overall system performance.
4Productivity
If simplified mirror images are used, then deployment speed is improved, but compatibility and functionality are reduced
Solution Approach 1:
The patent segments the mirror image into layers with different optimization levels. LALs (Layers Above Last Download Layer) are stored in NVM for fast access and contain critical application files, while LBLs (Layers Below Last Download Layer) are stored on hard drive. This segmentation allows the system to maintain full image functionality while optimizing deployment speed for the most critical layers, avoiding the compatibility issues associated with simplified images.
Data Source
AI summary
The present disclosure discloses a NVM-based method for performance acceleration of containers. The method comprises classifying each image layer of mirror images as either an LAL (Layer above LDL) or an LBL (Layer below LDL) during deployment of containers; storing the LALs into a non-volatile memory and selectively storing each said LBL into one of the non-volatile memory and a hard drive; acquiring hot image files required by the containers during startup and/or operation of the containers and storing the hot image files required by the containers into the non-volatile memory; and sorting the mirror images in terms of access frequency according to at least numbers of times of access to the hot image files so as to release the non-volatile memory currently occupied by the mirror image having the lowest access frequency when the non-volatile memory is short of storage space.