Container non-perception starting method and system

Abstract

The invention discloses a container non-perception starting method and system based on a serverless platform, and the method comprises the following steps: analyzing a user request through an interpreter, and transmitting the user request to a cold standby queue or a hot standby queue; the processor performs security authentication on the user request instructions of the cold standby queue and the hot standby queue, performs clustering analysis on the user requests passing the security authentication, and forwards the user request instructions subjected to the clustering analysis to the container manager; the container manager is responsible for creating a container, calling a trigger function and scheduling the container; and the resource monitor is responsible for creating, updating and closing containers in the hot standby pool, and comprises a combined container, a decomposed memory or a container of which the CPU exceeds a threshold value, and a periodically started and stopped container. Compared with the prior art, the method has the advantages that container initialization and dependent program package loading are accelerated, and resources are effectively saved.

Description

technical field [0001] The invention relates to a method and system for starting a container without perception, and belongs to the technical field of cloud computing. Background technique [0002] With the continuous development of cloud computing technology, serverless technology (serverless) has become an inevitable trend in the development of cloud computing. Enterprises only need to provide cloud providers with the cost of renting resources, and they can use resources to call services when user requests come. When there is no user request, it does not occupy any resources, and billing is based on the number of calls and the duration. Compared with the traditional online service model, serverless computing greatly reduces the user's cost of use, so that users do not need to pay attention to server configuration issues to simplify It simplifies development and provides better scalability than traditional online services. However, serverless adopts an on-demand structure,...

AI Technical Summary

Problems solved by technology

However, serverless adopts an on-demand structure, and idle function instances will be deleted during runtime. Therefore, when the function is called again, the platform will restart a new instance and redeploy the operating environment and software code, resulting in a large number of service calls. The delay of the function startup seriously affects the responsiveness of the serverless

[0003] Therefore, many researchers have proposed optimizations for the cold start problem of no service. For example, Oakes et al. tried to reduce the startup time of cloud functions by caching the packages required by the function in the working nodes, and proposed a shared package cache. When a cloud function is When assigned to a worker node, the cache checks whether the required package is cached. The disadvantage is that a large number of cached packages occupy service resources

Mohan et al. adopted a container pool strategy to reduce cold start delays. They pre-run empty containers so that new containers can be obtained directly from the container pool when the function is triggered without restarting the new container. However, this method uses a static container pool capacity , which will lead to waste of resources when there are few function user requests

Harter et al. studied the software packages installed when starting a function instance, so as to prioritize the necessary software packages to achieve rapid software deployment, and delay loading of non-essential software packages. Although this method can effectively reduce the instance startup delay, it still cannot avoid cold The emergence of startup, it is difficult to meet the requirements of fast response functions

Images