Hyperconvergence is a rapidly growing storage segment. More and more companies are choosing to move from traditional architecture to HCI. Do you want to know why? In the article, we will answer this question and tell you what tasks vSAN is suitable for on the example of a solution from VMware.

Converged Storage

Traditional architecture has proven itself over many years of its existence, it is reliable, practical, and versatile. All three of its components, namely servers, data storage systems (DSS), and storage networks, interact at high speeds and provide good fault tolerance, but its construction, maintenance, and scaling is a task of a long time and big money.

When the result is needed quickly, and in today’s business, cloud or otherwise, the need for computing is changing very dynamically – the “bricks” of converged systems become more convenient. Therefore, if you are building an infrastructure for resource-intensive business tasks or planning to scale it, pay attention to HCI.

VMware proposes to combine both compute and storage functions on one device – to deploy a converged software solution VMware vSAN. Such a converged storage system does not require special devices and data transmission networks, it works directly on the server and is able to perform the same task as an external storage system.

Principle Of Operation

VSAN uses the server’s disks to create virtual storage that is available to all nodes in the vSphere cluster. Programmatically, this works on the same hypervisor as the virtualization platform itself, in parallel with the computing part. Accordingly, both virtual storage and virtual servers work simultaneously on server – convergence. For fault tolerance, several nodes are needed, the disks of which will create a single storage space available even in the event of a failure on one of the servers.

Each host (cluster node, server) is part of the overall vSAN cluster and must have at least one disk group. The disk group consists of 1 cache disk (fast SSD) and 1 to 7 storage disk (regular HDD or also SSD). There can be up to 5 such groups on a server, therefore, the maximum amount of storage on one server will consist of 35 disk capacities, and the minimum one.

In order for a vSAN cluster to work, a minimum of 2 hosts is required, the disk groups of which will be mirrored to ensure continuous operation in case of failure. This configuration only works with the Mirroring fault tolerance method (similar to RAID1). In it, the Number of failures to tolerate parameter is equal to one, that is, data from one host is immediately synchronized on another. By adding another host and setting the FTT=2 parameter, the data will be replicated on two servers, which increases the degree of fault tolerance, but makes the use of disk space even less efficient. The maximum value of FTT=3 allows you to have data instances on 4 different hosts.

If you are using an All-Flash configuration, that is, only SSD drives are installed on the nodes, then you can use the Erasure Coding method (similar to RAID 5/6). Writing to disk groups is accompanied by the calculation of the corresponding parity blocks, which allow you to recover data in case of failure. So you can significantly save disk space but lose in speed and recovery time. The minimum number of hosts required for Erasure Coding to work is 4, but 5 to 7 is recommended.

The standard vSAN organization option uses hosts running ESXi hypervisors, but vSAN Appliance and Storage Appliance solutions are possible. With vSphere installed, you can activate vSAN without additional plug-ins and add-ons, and all management is performed from vCenter.

vSAN Practice Features

Successful implementation of vSAN depends on many factors, but there are basic requirements and features that you need to know in order to avoid mistakes.

  • You should not use the maximum number of disks for storage in a disk group. It’s better to have two disk groups of three disks than one with six. This is more expensive as it requires an additional cache SSD but reduces the size of the fault domain. If a cache drive fails, three drives will rebuild faster than six. And the speed of parallel operation of two disk groups is higher than one, especially if they are hung on different controllers. Disks are also better to use not the maximum volume (for example, 6TB, but 1-2 TB).
  • Do not set the value of the FailuresToTolerate parameter to less than one, as this will keep the cluster running without redundancy, and you risk losing data in the event of a disaster.
  • vSAN doesn’t have any special hardware requirements, you can use standard x86 servers, but it’s best to follow VMware’s list of recommended hardware. It is better to write the hypervisor to the built-in flash drive, the SSD for the cache to the PCIe slot, and the storage drives to the standard controller ports.
  • Since no dedicated network is used for data transfer, but ordinary Ethernet, its bandwidth must be at least 10GB.
  • In a hybrid configuration, the SSD caches both reads and writes, while All-Flash caches only write. Therefore, it is better to use just such a configuration, since 100% cache under the write buffer gives a strong advantage in the speed of processing the incoming stream of I / O operations. The higher the amount of SSD cache, the better the performance. If you use All-Flash, then no one bothers you to use an SSD drive with better characteristics than the rest for the cache.
  • Speaking about vSAN, easy scaling is often emphasized. Indeed, in order to add computing power or additional disk space, an additional node must be connected to the network. And if only disks are needed, then it will not work to increase the capacity of virtual storage on the go, simply by adding or replacing it with larger disks. It is necessary to calculate the ratio of storage and computing power in advance.
  • You can activate and work with vSAN immediately if you have vSphere, but do not forget that the technology is licensed separately by the number of processors. Therefore, the more nodes in your cluster, the more expensive virtual storage is. Therefore, you will need to calculate the budget in advance.
  • Conclusion

    Using vSAN reduces costs, and provides flexible scalability and easy management of a single VMware ecosystem. HCI solutions are at the peak of popularity, this is due to their versatility and speed of implementation. Businesses require instant response to change and converged solutions like vSAN deliver fast reliable, and productive results.