Browse Prior Art Database

An Efficient Massive Deploy System on Cloud using a Distributed Protocol

IP.com Disclosure Number: IPCOM000244139D
Publication Date: 2015-Nov-13
Document File: 5 page(s) / 174K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system to perform massive deployments of virtual machines by sharing common information and using a distributed protocol and the same image seed on a cloud system. The system combines a Copy-on-Write (COW) with a distributed protocol to spawn virtual machines.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 5

An Efficient Massive Deploy System on Cloud using a Distributed Protocol

The massive deployment of virtual machines (VMs) often results in poor performance because the existing solutions spawn instances serially or even in parallel, but not in an efficient manner. Massive deployment consumes a considerable amount of system resources (e.g., network, memory, central processing unit (CPU), storage, etc.). A novel system is needed to manage these resources in order to achieve better performance and utilization efficiency.

Scalability is not good with the deployment of greater than 100 (N > 100) virtual machines. The process retrieves the same image from the image service for each virtual machine, which hurts performance and slows the VM spawn. (Figure 1)

Figure 1: Common spawning scenario for a cloud provider

This is a cloud provider that has N servers and a unique image server provider. The goal in this scenario is to spawn a certain number of VMs in each server. The common approach to execute this task is:

1


Page 02 of 5

for each server in SEVER_POOL for each VM to be Created if the VM_IMAGE is not in the HOST_CACHE get _image from IMAGE_PROVIDER

else

get_image from HOST_CACHE

Image files are large, often tens of GigaBytes (GB) in size. When a system receives a request for an image, it must transfer the entire file to the hosting physical server for instantiation. For example, to create 100 VMs, the system attempts to locate the VM image in the host cache. If it is not there, then the system copies the entire image from the image service provider.

The majority of time necessary to provision a request for a VM in the cloud is spent delivering an image from an image repository , typically due to bandwidth-limited connectivity between sites (and the large size of the file). Therefore, accessing a repository over the Internet to fetch a GB-sized file can significantly decelerate server- provisioning speed.

The novel contribution is a system to perform massive deployments of virtual machines by sharing common information and using a distributed protocol. It manages a massive deployment of VMs with the same image seed on a cloud system. The system combines a Copy-on-Write (COW) with a distributed protocol to spawn virtual machines. It enables a massive deployment using a distributed method. In addition, the system applies ahead-of-time deployment (i.e. propagate a common change across the guests) and a neighbor cache from a torrent-like protocol.

The idea is that most operati...