VMware vSAN Getting Started Summary

1. Background

1. Briefly introduce the shared storage background of vsphere

An important function in vSphere is virtual machine encapsulation. A virtual machine exists in the form of a file and can be copied at will, such as a .vmx virtual machine configuration file, a vmdk data file, etc.

vSphere also has the concept of clusters. A cluster is considered a resource pool. With many advanced features of vSphere, services can be run on any host in the cluster without worrying about single host failures.

As shown in the figure below, vSphere's fault recovery mechanism HA can migrate virtual machines on a failed host to other hosts for operation.

However, this feature has a prerequisite of shared storage. One storage can be connected by multiple servers at the same time and read data at the same time. If any node fails, the data will not be affected, and other servers can use this data to quickly resume business.

The dual-head redundant storage is as follows:

2. Simple comparison between shared storage and distributed storage

Local storage has always been known for its high reliability, good stability, and rich functions. However, it has poor storage scalability, high prices, and difficulty in data connectivity. It is easy to form data islands, resulting in high management and maintenance costs for data centers.

Distributed storage: Data is stored on multiple independent devices on the network. It is usually connected by standard x86 servers and networks, and related storage software, such as vSAN, is run on them. The system provides storage services to the outside world as a whole.

2. Introduction to vSAN

1. Official Introduction

What is vSAN can be understood as software-defined distributed storage based on server clusters.

The storage management program is implemented by software. Only software can be developed flexibly and quickly to adapt to the various storage needs of enterprises. Distributed storage can disperse data and IO access to multiple nodes, allowing the capacity and performance of the entire storage system to increase linearly with the increase of nodes.

The diagram is as follows:

VMware vSAN aggregates local or direct-attached data storage devices and creates a single storage pool that is shared among all hosts in the vSAN cluster. vSAN eliminates the need for external shared storage and simplifies storage configuration and virtual machine configuration.

vSAN is a distributed software layer included in the ESXi hypervisor that is fully integrated with VMware vSphere. vSAN supports vSphere features that require shared storage, such as High Availability (HA), vMotion, and Distributed Resource Scheduler (DRS). VM Storage Policies enable you to define virtual machine storage requirements and capabilities.

Each host in a vSAN cluster can contribute storage to the cluster. These storage devices are combined together to create a single vSAN datastore. A hybrid vSAN cluster uses flash devices as the cache tier and magnetic drives as the capacity tier. An all-flash vSAN cluster uses flash devices for both the cache and capacity tiers. The architecture creates flash-optimized, resilient shared data storage designed specifically for virtual environments.

2. vSAN Concept

A key element of software-defined storage is storage policy-based management (SPBM), a feature of vSAN. Through SPBM and the vsphere API, the underlying storage technology is presented to administrators as an abstract pool of storage space with multiple functions for deployment.

You can configure a hybrid cluster or an all-flash cluster. In a hybrid cluster, flash is used for cache and disk is used for capacity. In an all-flash cluster, flash is used for both cache and capacity. (The combination of cache and capacity makes storage performance better, and all-flash is definitely faster)

Local storage devices, which must include at least one flash cache device and one capacity device

Each node needs to provide storage resources, access resources and management programs

Typically requires a 10 Gigabit Ethernet switch

The devices on the host will form one or more disk groups, each of which contains a flash cache device and one or more persistent storage capacity devices. Multiple disk groups can be configured on each host.

3. vSAN functions and features

characteristic:

• Built into the vsphere kernel, vSAN is implemented within the vSphere kernel and seamlessly integrates with vSphere to help optimize performance and reliability.
• Read/write (I/O) caching: vSAN uses built-in cache in server-side flash memory to speed up read/write disk I/O and minimize storage latency.
• Built-in fault protection, which uses distributed RAID and cache mirroring to ensure that data is never lost in the event of a disk, host or network failure
• Non-disruptive capacity scalability allows you to easily and non-disruptively expand vSAN data capacity by adding hosts to the cluster or adding disks to the hosts.
• With VM-centric policy-based management, the solution associates storage requirements with individual VMs or virtual disks using policy statements that are automatically converted into system configurations. This approach allows IT to instantly provision storage to strictly adhere to service-level agreements (SLAs).
• With self-adjusting storage and dynamic storage load balancing, vSAN automatically and non-interruptively maintains the storage capacity, performance, and availability levels specified for each virtual machine. The technology can interoperate with VMware vSphere Distributed Resource Scheduler to achieve end-to-end technology and storage load balancing.
• Integrated with vSphere Data Services, the solution leverages vSphere snapshots, clones, VMware vSphere Data Protection, and vSphere Replication to provide data protection, backup, rapid cloning, and data transfer across clusters or sites for disaster recovery.
• Integration with vSphere web client
• Wide hardware compatibility
• Interoperable with Horizon View and VMware vCenter Site Recovery Manager, the solution can be deployed with Horizon View in a virtual desktop infrastructure (VDI) environment and with vCenter Site Recovery Manager in a disaster recovery environment

Function:

• Shared storage support. vSAN supports VMware features that require shared storage, such as HA, vMotion, and DRS. For example, if a host is overloaded, DRS can migrate virtual machines to other hosts in the cluster.
• Just a Bunch of Disks (JBOB) supports the use of JBOB in a blade server environment. If your cluster includes blade servers, you can use JBOB storage connected to the blade servers to expand data storage capacity.
• All-flash and hybrid configurations
• Disk format: vSAN 6.6 supports disk virtual file format 5.0, which provides highly scalable snapshot and clone management support per vSAN cluster.
• Fault domains. vSAN supports configuring fault domains. When vSAN clusters multiple racks or blade server chassis in a data center, it can prevent the host from experiencing a rack or chassis failure.
• Stretched clusters, vSAN supports stretched clusters, which can span two address locations
• vSAN Health Service, which contains a number of preconfigured health check tests that can perform health and troubleshooting and diagnose the cause of problems with cluster components and identify any potential risks
• vSAN Performance Service, which includes statistical charts for monitoring IOPS, throughput, latency, and congestion. Can monitor vSAN cluster, host, disk group, disk and virtual machine performance
• Integration with vSphere storage features, typically vSphere data management features for VMFS and NFS storage, including snapshots, linked clones, vSphere Replication, and vSphere APIs for data protection
• The VM storage policy and domain VM storage policy can be used together to support VM-centric storage management methods, such as allocation, which will automatically reduce the vSAN default storage policy to the VM.
• Rapid provisioning: vSAN can quickly provision storage in the vCenter server during creation and deployment.

4. Some vSAN terms and terminology

Pooling

• Connections within a single machine
• Connections between servers

Fault Domains

In order to ensure that data is not lost, the storage location of the data has certain requirements. The same data of the same virtual machine must be stored on different hosts.

If one server fails, an arbitration mechanism is needed to ensure that only one copy of the data is active and up-to-date, otherwise conflicts will occur.

This is the simplest architecture of vSAN, which only allows one host failure.

The following is a simple diagram of vSAN fault domains. In vSAN, there is a term called FTT (Fault to tolerance), which refers to the maximum number of hosts that can fail simultaneously. FTT determines the level of data protection for virtual machines and also determines the minimum number of hosts required for a cluster. The number of hosts in a cluster is 2n+1, where n = the value of FTT.

Differentiated Services

You can use storage strategies to differentiate different services for different objects.

For example:

• Virtual machine 1 sets storage policy A (FTT=1, no cache reserved, IOPS limited to 100)
• Virtual machine 2 sets storage policy B (FTT=2, reserve 10% of SSD cache, and do not limit IOPS)

Disk Group

Disk groups are units of physical storage capacity on hosts and groups of physical devices that provide performance and capacity to the vSAN cluster. On each ESXi host that provides its local devices to the vSAN cluster, the devices are organized in disk groups.

Each disk group must have a flash cache device and one or more capacity devices

Each disk group must have at least one SSD and 1 to 7 HDDs. Each host must not have more than 5 disk groups.

Devices used for caching cannot be shared between disk groups or used for other purposes. A single cache device must be dedicated to a single disk group. In a hybrid cluster, flash devices are used for the cache layer and magnetic disks are used for the storage capacity layer.

It is recommended to use multiple disk groups to disperse data. For example, if an SSD fails, the data in all disk groups will be inaccessible. Dispersing data can effectively reduce the impact of this failure.

All Capacities

• All capacity refers to the physical capacity used by one or more virtual machines at any time. The used capacity is determined by many factors, including the capacity used by VMDK, including replicas, etc. When caching, the capacity used for including replicas is not considered.

Object-based storage

1) vSAN stores and manages data in the form of flexible data containers, that is, objects. The data and metadata of an object are distributed across logical volumes in the cluster. For example, each VMDK and each snapshot is also an object. When provisioning a virtual machine on the vSAN data store, vSAN creates a set of objects consisting of multiple components for each virtual disk.

2) Based on the assigned VM storage policy, vSAN provisions and manages each object individually and may also create a RAID configuration for each object

3) When vSAN creates objects for virtual disks and determines how to distribute objects across the cluster, it considers the following factors:

A. vSAN confirms that the virtual disk requirements have been applied according to the specified VM storage policy settings.

B. vSAN confirms that the correct cluster resources are used for provisioning. For example, vSAN determines the number of replicas to create based on the protection class, the performance policy determines the amount of Flash Read Cache to allocate for each replica and the number of stripes to create, as well as where to place each replica in the cluster.

C. vSAN continuously monitors and protects the policy compliance status of virtual disks. If any non-compliant policy status is found, troubleshooting must be performed and the root cause must be resolved.

D. If necessary, you can edit the VM storage policy settings. Changing the storage policy settings will not affect VM access. When you change the VM storage policy settings, vSAN may start the object recreation process and then resynchronize the objects.

E. vSAN confirms that the required protection components (such as images and witnesses) are located on different hosts or fault domains

vSAN Datastore

1) After enabling vSAN on a cluster, a separate vSAN datastore is created. A single vSAN datastore can provide different service levels for each VM or each VM disk.

1) vSAN data storage specific features need to be considered

A. vSAN provides a single vSAN datastore that is accessible to all hosts in the cluster (regardless of whether they have devices). Each host can mount any other datastore, including Virtual Volume, VMFS, or NFS.

B. You can use Storage vMotion to move virtual machines between vSAN datastores, NFS, and VMFS datastores.

C. Disks and flash devices used only for capacity can provide data storage capacity. Devices used for flash cache are not counted as part of data storage.

D. In automatic disk claiming mode, when you add a host with capacity to the vSAN cluster, or add a capacity device to any cluster member, the vSAN datastore size grows dynamically.

Objects and components

1) Each object consists of a set of components, such as .vmx, log files, vmdk, snapshot incremental description files, etc.

2) vSAN datastore protects the following object types:

A. Virtual machine home command space

All virtual machine configuration files are stored in the virtual machine home directory, such as .vmx, log files, vmdk (the vmdk disk description file fvmdk-lag is the disk file), snapshot incremental description files, etc.

B. VMDK

A virtual machine disk, or .vmdk file, stores the contents of a virtual machine's hard drive.

C. Virtual machine exchange object

Create -SWDK file when powering on the VM

D. Snapshot incremental VMDK

Created when a virtual machine snapshot is taken

3) vSAN components are blocks of objects that are distributed across multiple hosts in a machine to allow for multiple failures to occur simultaneously and meet performance requirements

4) vSAN uses a distributed RAID architecture to distribute data across the cluster

5) The distribution of components mainly adopts two technologies:

• Striping (RAID 0)
• Mirroring (RAID 1)

6) How many component replicas are created is determined based on the object strategy definition

Virtual machine compliance status: Compliant and Non-Compliant

A virtual machine is considered non-compliant when more than one object fails to meet the requirements of the assigned storage policy. For example, if a mirror copy is inaccessible, the status may become non-compliant. The status of the virtual machine is compliant when the requirements defined in the virtual machine domain storage policy are met. In the Physical Disk Placement tab of the Virtual Disks page, you can determine the compliance status of a virtual machine object.

Component status: Degraded and non-existent status

Degraded. A component is in a degraded state if a permanent fault is detected and it is determined that it cannot be restored to its original working state.

Does not exist. vSAN detects a temporary component failure, where the component and all its data can be recovered and vSAN can restore its original state. The component is in the non-existent state. This may occur when you restart the host or unplug the device from the vSAN host. After waiting for 60 minutes, vSAN will rebuild the non-existent component.

Object status: normal and abnormal

Depending on the type and number of failures that occur in the cluster, an object can be in one of the following states:

Normal, when at least the RAID 1 mirror is available or the minimum required number of data segments is available, the object is considered to be in normal state.

Unhealthy. If no complete mirror is available, or if a RAID5 or RAID6 object does not meet the minimum number of segments, the object is considered unhealthy. If the object has less than 50% of votes, the object is unhealthy. If there are multiple failures in the cluster, the object may become unhealthy, affecting the availability of the virtual machine.

prove

A component that contains only metadata, without any actual application data, and can be used as a te breaker when it is necessary to determine the availability of a healthy data storage component if a potential failure is detected. If you use disk format 1.0, the witness will take up 2MB of space in the vSAN datastore to store the metadata. If you use 2.0 or higher, it will take up 4MB of space.

6.0 and higher maintain arbitration through a non-voting system, where each component may use multiple votes to determine the availability of an object. If more than 50% of the votes that make up a virtual machine storage object are lost, the vSAN datastore will no longer be able to access the object. Inaccessible objects may affect the availability of related virtual machines.

Striping

It is a technology that automatically balances the I/O load to multiple physical disks. Striping technology is to divide a continuous piece of data into many small parts and store them on different disks. This allows multiple processes to access different parts of the data at the same time without causing disk rereading. Moreover, when such data needs to be accessed sequentially, the maximum I/O parallel capability can be obtained, thereby achieving very good performance.

The following is an example:

Number of disk stripes per object

The number of hard disks across which each copy of the stored object spans

Storage Policy Based Management (SPBM)

When you use vSAN, you can define the storage requirements of your virtual machines, such as performance and availability, in the form of policies. vSAN ensures that at least one virtual machine storage policy is assigned to a virtual machine that is deployed to a vSAN datastore. When you understand the storage requirements of your virtual machines, you can define a storage policy and assign it to the virtual machine. If you do not apply a storage policy when you deploy a virtual machine, vSAN automatically assigns the default vSAN policy, which has a primary level of failures to tolerate of 1, one disk stripe per object, and one thin provisioned virtual disk. For best results, you should define your own virtual machine storage policy, even if your policy requirements are the same as those defined in the default storage policy.

Ruby vSphere Console (RVC)

RVC, provides a command-line interface for managing and troubleshooting vSAN clusters. RVC provides cluster-wide views instead of the host-centric views provided by esxcli. It is bundled with vcsa and does not require separate installation.

vSphere PowerCLI

Added command-line scripting support for vSAN to help automate configuration and management tasks

vSAN Observer

It is a web-based tool available on RVCS for performance analysis and vSAN cluster monitoring. vSAN Observer can be used to view performance statistics of the capacity layer, statistics of physical disk groups, current CPU load, usage of vSAN memory pools, and distribution of memory objects across the vSAN cluster.

5. vSAN Core Concept

• Improve the utilization of existing resources
• Reduce operation and maintenance costs
• Improve hardware and application availability
• Achieving operational flexibility
• Improve desktop manageability and security

6. Main functions implemented by vsphere web client

• vMotion, DRS, HA and other advanced features are based on this technology
• DRS dynamic resource allocation technology enables virtual machines running on multiple physical servers to achieve load balancing. When it is detected that the load of a server is approaching the threshold, online migration can be performed using vMotion technology.
• DPM distributed power management can be combined with the DRS function to concentrate multiple virtual machines with low loads on a few physical servers, shut down useless ones, and save electricity. When the load increases, power on the unused physical servers and achieve load balancing through DRS.
• Storage vMotion storage online migration technology allows the running virtual machine storage resources to be migrated from one storage device to another storage device
• Data protection function, which can back up virtual machines
• High availability: when a physical server fails, the affected virtual machines can be migrated to other servers.
• FT fault tolerance, running two identical virtual machines on different physical servers, provides the highest level of redundancy. Even if one physical server fails, the business operation will not be interrupted.
• replication, which can copy all images of the virtual machine to another site for disaster recovery and ensure data security

Reference: https://docs.vmware.com/cn/VMware-vSphere/6.5/com.vmware.vsphere.virtualsan.doc/GUID-18F531E9-FF08-49F5-9879-8E46583D4C70.html

Reference: https://new.qq.com/omn/20180717/20180717G0AUBL.html

This is the end of this article about the introduction to VMware vSAN. For more relevant VMware vSAN content, please search 123WORDPRESS.COM's previous articles or continue to browse the following related articles. I hope everyone will support 123WORDPRESS.COM in the future!