Shulou(Shulou.com)06/03 Report--

What I will introduce to you today is the structure and principle of a common storage device-Vsan, which is relatively technical. VSAN is a distributed storage architecture that can be extended and used based on the vSphere kernel. The hard disk and flash memory stored in the vSphere cluster host build a VSAN storage layer, which is managed and controlled through storage, and finally forms a shared storage layer.

With the rapid development of computer network, the storage structure of vsan is constantly updated. The underlying storage in the traditional storage management mechanism does not understand virtualization and file system, and the new generation of storage management mechanism will be updated to object-based storage system, virtual data storage and distributed storage. The following figure is a schematic diagram of the storage structure and storage management mechanism of vsan.

The VSAN data store is an object store that is presented to the vSphere host as a file system, and the object storage service mounts the volume from each host in the VSAN-enabled cluster. By presenting them as a single distributed shared data store visible on all nodes, VSAN simplifies storage configuration. For virtual machines, there is only one data store, which comes from the storage space on each vSphere host in the VSAN cluster and is configured through disk groups to store all virtual machine files in a separate storage entity. This method of data storage is relatively safe, but when there is a failure of the flash drive or capacity disk, the data will be transferred to other nodes, and other failures may occur during the transfer. Here is a case of vsan storage data recovery that has been successfully restored recently.

The failure is a VSAN cluster composed of four dell servers, and each server has two disk groups. One disk group is a SSD hard disk with 5 SAS hard drives, SSD does flash memory, and SAS does capacity disk. The capacity disk in a disk group on one node fails offline. At this time, VSAN carries out data reconstruction and migration. At a very unfortunate time, data migration is not completed because of a power outage. At the time of an incoming call, the capacity disk in the other disk group was also offline by two failures, resulting in a failure of the entire data storage. The VSAN management console could log in, but all virtual machines could not be accessed.

First, make a read-only image of all the hard drives of the four nodes, including SSD flash drive and SAS capacity disk, and three hard disks offline due to failure. After the backup is completed, restore all the original disks to the server, and begin to analyze the underlying data storage structure of the image files to confirm the distribution information of the virtual machine's hard disk, because the existing virtualization programs do not restore the virtual machine for the VSAN architecture. When analyzing the underlying data storage structure, engineers in North Asia are also doing corresponding program development to test the accuracy of data distribution information.

Analyze the two disk groups on each node separately, take a look at the corresponding relationship between the flash hard disk and the capacity disk in the disk group, each hard disk has a unique identification to correspond to each other, and interpret the hard disk ID information in the disk group according to the ID information of the hard disk.

1. Get the UUID of the disk and UUID of the disk group on each hard disk

2. Obtain the component information of the capacity disk according to the component information of the capacity disk in each disk group.

3. Extract the component bitmap according to the MAP position of the component recorded in the component information.

4. extract component data and cache data according to the bitmap of the component

5. Get the object and the order of the component according to the description information of the component, and merge the component into an object.

6. Extract data according to the object.

The object can also be regarded as a volume, or it can be understood as a logical volume. Each VSAN object that exists on the data store is composed of multiple components, which are distributed in the disk groups configured on the cluster host. In the process of recovery, the information extraction of the components is a key step, because the components are an important part of each object, and the failure components are rarely damaged. The restored virtual machine can start normally, this fault resolution takes a long time to analyze the corresponding relationship between component bitmap and disk, but in the end, all the technical problems are solved, and the restored virtual machine starts normally, which satisfactorily solves the recovery of data loss caused by VSAN failure.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.