Storage System Design Scheme in Virtualization Construction

Si Zhen-yu

Network Information Center, Northeast Agricultural University, Harbin 150030, China

Introduction

Virtualization has become one of the main promoters of the modern data center, it can improve the resource utilization, system support options and the response ability, while cloud computing technology further promote its development, which makes it possible to provide on-demand configuration and scalability for the users, therefore, virtualization can be regarded as the basis of cloud computing. In computer technology,virtualization refers to abstract the physical resources,such as servers, network, and memory, and presents them after the conversions, so that the users can apply these resources in a better way than the original conf i guration. The new virtual parts of these resources are not restricted by the erection way of the existing resources and geographical or physical configuration. The virtualized resources generally include computing power and data storage (Bowker, 2011; Liao,2006).

Storage system is the most important part of virtualization, it is said that storage system is the key factor of virtualization, because of its high availability and continuity. In virtualization construction, storage systems store various structured data such as databases produced in virtualization environment and a variety of text, images, sound, video, and other unstructured data files, as well as various environmental documents of the virtual machine. Storage systems not only store and backup all kinds of fi les, but also deal with a variety of data access requests. And the storage response ability to access directly affects the access speed and integrity of the entire virtualized environment. In the process of data access, the data need to provide access request through the network and then be processed by the physical host memory and CPU, when the users access to the data in the virtualized environment, the response ability provided by storage is far less than the physical host CPU and the memory capacity in the environment in the entire process of data fl ow. Network backbone transmission speed unit is Gigabit; memory bandwidth is generally six GBps; CPU processing power is even higher than the memory bandwidth. While memory bandwidth is generally just one MBps, it is lower than others. As can be seen, the storage is the bottleneck of the whole environment, we must optimize the existing storage systems in order to improve the entire virtual environment (Johnson, 2008; Gulati, 2010).

A good system should not only be able to provide fast and efficient storage service, but also has a perfect backup mechanism to deal with all kinds of accidents that may occur in the process of system running.

At present, in the construction process of virtualization, storage areas mainly use the centralized storage way at home and abroad, which unif i es storage devices to store fi les in the virtual environment. It has achieved a rapid development mainly in the aspect of improving hardware storage performance, and produces a variety of advanced hardware storage technologies, such as the use of the solid-state storage media, the use of tiered storage, and reduplication data deletion technology,but the use of new technologies brings about increased construction cost, so there is a certain one-sidedness that singly improve the performance of storage hardware instead of increasing the entire storage system.If we take the virtualization construction cost into consideration, the feasibility will be greatly affected in the case of limited investment.

Based on the current program of virtualization construction and implementation, this paper studied the defects of current centralized storage solution, and established a system storage solution program that fi tted for the virtualization construction by optimizing infrastructure, improving service, integrating resources,and promoting application under the conditions of a certain cost control. The new storage system could be guaranted to meet the requirements of storage security and continuity, and the storage and backup of virtual machine's structured and unstructured files under the virtualization environment.

Storage System Design in Virtualization Construction

Defects of current centralized storage solution

The storage system design of virtualization construction is mainly centralized storage, and then set up a backup disaster recovery equipment, centralized storage is generally as much as possible with more advanced storage technologies, such as automated tiered storage, reduplication data deletion, and highspeed solid state storage devices (Fig. 1).

Fig. 1 Virtualization centralized storage structure

As shown in Fig. 1, there are several virtual machines are running and provide services above the virtual host, they provide a variety of services, and all point to the centralized storage, the centralized storage platforms generally use iSCSI, SAN or NAS to storage. Virtual host does not have storage function,when the virtual host fails, all the data are not in the local, but in the storage platform, so we can move virtual machines on other physical host quickly to restore and run, so as to achieve high availability,high persistence, and high security. Backup storage platform can provide backup of centralized storage platform to maintain data recoverability.

The virtual machine in the virtual architecture system is actually packaged into a fi le and a number of related environment configuration files, which are stored in the fi le system of the SAN memory array and can be accessed by the virtual machines in different services, in this way, we can eliminate the cases of unavailable application system resulted by single point failure (Han, 2011). For instance, a virtual machine can provide web access services, numbers of website file directories are stored in this machine, when the virtual machine works, the website file directories and other files stored in it are integrated into a large capacity and special fi le, and it will be a great loss if there is an accident. This design is reasonable but not comprehensive, because it saves the virtual machine fi les, which are one or more large fi les, including the structured and unstructured documents generated by the running virtual machine, if there is a failure in this large fi le, the useful data will be very diff i cult to recover.

Hardware and software environments of new storage system

The new storage system is designed based on the existing centralized storage system architecture, in addition to the existing network and storage equipment,a NAS storage equipment is combined, which must satisfy the condition that every virtual machine can access to it and have written to the storage rights,what's more, in order to ensure the system safety, it is better to have a independent network address which is different from the management network and the centralized storage. The new storage system design is shown in Fig. 2.

Fig. 2 New storage system architecture design

Design of new storage system

Design ideas: in addition to storing the virtual machine environment files, the new storage system store and backup the fi les in the virtual machine.

Solution: to solve the problem, we need to build a shared storage platform specifically for the virtual machine to access, which stores fi les by NAS or iSCSI way. When the virtual machine works, the structured and unstructured data files produced by it will be stored in the shared storage platform and become real fi les from virtual ones. Thus, the existing centralized storage platform and the shared storage platform form a more complete storage platform, the single large fi le of the virtual machine environment is stored in the centralized storage platform, and all kinds of structured and unstructured fi les produced by the running virtual machine can be stored in the shared storage platform,based on it, all data can be saved and restored.

Through the above structure, it can be seen that in the process of virtualization construction, the new storage system can not only intently store the separate fi les of the virtual machine, but also store all fi les by the NAS storage device, which has high availability and integrity.

Conclusions

By the way of building a shared storage platform to store various files in the virtual machine, we converted these fi les into physical ones and provided a comprehensive storage system, which supplemented and improved the centralized storage platform. After the implementation of virtualization, all files in the virtual machine or produced by it could be found on the storage platform, which provided a good protection for virtualization construction. This storage system could provide better base and powerful reference for the future virtualization construction and even the cloud computing construction and implementation.

Bowker D. 2011. Animal record management using an embedded RFID-based system. The 118th ASEE Annual Conference and Exposition.

Gulati A. 2010. Modeling workloads and devices for load balancing in virtualized environments. The 2nd Workshop on Hot Topics in Measurement and Modeling of Computer Systems.

Han Q, Bao W. 2011. A halal and quality attributes driven animal products formal producing system based on HQESPNM. The 4th IFIP International Conference on Computer and Computing Technologies in Agriculture and the 4th Symposium on Development of Rural Information.

Johnson A K. 2008. Setting the farm animal welfare science in North America. The 8th International Livestock Environment Symposium.Iguassu Falls, Brazil.

Liao A H. 2006. A 3D registration method for US/PET multi-modality small animal tumor imaging. IEEE International Ultrasonics Symposium.