Similar to the concept of Virtualization of Servers or Virtualization of Network Services, the Virtualization of storage comes to solve the complexity of the process of provisioning and administration of storage resources, which normally requires a manual process, prone to failures, which can become demanding and also critical for today’s organizations.
When we think about acquiring a storage solution such as those offered by EMC, Dell, NetApp, HP, etc., we see that the prices change according to the brand and the advanced storage options, not so much because of the hardware functionalities, which is almost always similar in all brands. In some cases, these virtues are provided mainly in the administration software included by the manufacturer, software that offers very little, or no abstraction capacity with respect to hardware. On the other hand, it is clear that when buying a storage solution from a particular manufacturer, we are paying an additional portion just for the fact of buying a brand that we are incorporating. This means that the person who manages the storage infrastructure must have the necessary skills to manage the software included by the manufacturer, which in turn becomes more difficult to administer and limits the competent professionals due to their lack of experience in said product as storage solutions and different manufacturers are added to the same infrastructure.
Talking about Software Defined Storage, or SDS, gives us the possibility of abstracting from the layer of hardware that forms our storage infrastructure, which in other words means that we can acquire any brand of hardware for the infrastructure of storage (the concept of Commodity), and manage it centrally through a software solution. Isolating storage infrastructure management software allows customers to manage various storage types and brands from a single software interface, simplifying management.
This administration software will provide both advanced functions such as Distributed Resource Scheduler, Storage DRS, High Availability, Snapshots, Streched Clusters, Replication, Thin Provisioning, Deduplication, etc., which are not dependent on storage hardware, and give the client the freedom to add new storage arrays, of any brand or type, when necessary.
The features of the Defined Storage SDS Software include:
- Hardware becomes a commodity with separation of storage in the software layer
- Separation of the hardware layer that forms the storage infrastructure
- Automated storage according to established standards
VSAN allows you to use the local disks of each ESXi host to provide shared storage in a group of servers (cluster).
The storage defined by the software is the technology that comes from being integrated into the ‘Data Center defined by the software’, previously with server virtualization, network services virtualization, security and now virtualization of the layer storage. The strategy seeks greater simplicity, efficiency and savings in storage costs, similar to how it was done with vSphere and NSX.
Exceeding virtualization standards and transforming data centers made up of hardware in software-defined environments gives us a vision of the spirit of virtualization: “A cloud, any application, any device.”
Greater benefits of vSAN
VMware vSAN can be used as a hybrid storage infrastructure or fully flash, depending on the preference and according to the needs. vSAN does not need the installation of additional software, since it is completely included in the hypervisor and activated with a few clicks. It can also be managed through the vSphere web client and integrated with the VMware platform. With this integration, the highest levels of performance will be obtained without causing an additional overload in the CPU capacity.
Advantages of vSAN
- Simplified and accelerated storage administration and provisioning
- Optimized flash storage architecture integrated in the hypervisor
- 90K IOPS per host (host) when using All-flash
- Accepts from 2 and up to 64 hosts per server group
- 50% reduction of the total cost of ownership (TCO)
- Automatic adjustment that balances storage resources
- Up to 8 PB of storage capacity can be achieved
Server virtualization is the platform that most prevails in data centers but it brings with it new requirements for the storage part. The storage of today that serves virtual infrastructures must serve different IOPS of The virtualization of servers is the platform that prevails in the data centers but it brings with it new requirements for the storage part. Today’s storage serving virtual infrastructures should serve different IOPS of different applications and workloads such as VDI. With several virtual machines on a dynamic platform you need to add visibility and control of the storage layer to the virtualization layer what is needed is a new approach to storage and that is achieved with VMware vSAN.
vSAN uses a hyperconverged storage architecture that enables compute and storage resources that are delivered through a virtual platform in common. Virtual SAN can be designed in a hybrid way or as the one known as All-Flash. In the hybrid architecture Virtual SAN takes magnetic disks and flash drives which are inside the servers to create a shared and distributed Data store. The flash discs offer us the temporary memory (caching) to improve the performance and in the magnetic discs is where the data will be stored. In the All-Flash architecture, only flash disks are used that provide us with temporary memory and persistent data services which give us an extremely high and predictable performance.
As Virtual SAN is already included in the vSphere core, it manages to optimize the IOPS of the data and minimize the impact on the CPU. The distributed design of the hypervisor gives the best performance to the applications. Virtual SAN is very simple and is part of the hypervisor, no additional software is required for installation, it is enabled only with two clicks. It offers us a very easy method that allows us a very simple daily management. What we must do is create storage standards, assign them to virtual machines and vSAN will do the rest automatically. We no longer have to deal with RAID’s or LUN’s anymore; also when we need to modify the rules we can do it on the fly, you just have to make the changes and vSAN will adapt them to the virtual machines automatically.
Virtual SAN is also very scalable, when you need to expand capacity or performance we just need to add host to the group of servers or disks to the host, without needing any interruption. The infrastructure grows as we need it, with the investment diluted throughout the growth period, depending on the requirements.
When we talk about prices for performance, vSAN offers the best value for the virtual infrastructure, since instead of having to invest in large, robust, oversized and mainly very expensive boxes, individual discs can be bought, which allows us to reduce the Capex to the use hardware at a lower cost and grow more tightly buying only what is needed when needed.
vSAN is radically simplified storage for VMware infrastructures. You get high performance, scalability, and a reduced TCO.
Among the requirements for each host we can mention that they include at least 1 SSD disk (PCI Express SSD / SAS / SATA /) and one HDD disk (SAS / SATA / NL-SAS); Adding additional disks or hosts can increase capacity and performance over time. It is also important to mention that in a group of vSAN servers not all hosts must add resources because there may also be hosts that only consume resources from it.
Each host that contributes storage to the VSAN server group requires a disk controller. This driver can be a SAS or SATA HBA, or a RAID controller. In case of using a RAID controller, it must work in one of two possible ways:
- Pass-through mode: Also known as JBOD or HBA mode, it is the preferred configuration for VSAN, since it allows VSAN to manage the RAID configuration based on defined storage rules.
- RAID 0 mode
A group of VSAN servers must have a minimum of 2 ESXi hosts that contribute their local disks, and a maximum of 64 hosts. The sizing of VSAN depends on the particular requirements of each client, both current and future, also considering that it is possible to easily scale by adding new ESXi hosts, or adding disks to existing hosts.
It is recommended that the capacity (size) of the SSDs be at least 10% of the capacity of the HDD disks in the server cluster to ensure performance. While greater the capacity of the SSD disks in a host, the cache can support a larger number of I / O operations (input / output), which will make it possible to achieve better results in terms of performance.
It should be remembered that SSDs in a group of VSAN servers will be used as temporary read / write memory, and their capacity will not be included in the shared Data store.