Many vendors design their hyper-converged products so that the initial configuration process is performed through...
a Web browser. This means the storage administrator deploying a hyper-converged architecture will typically have to open a Web browser and enter a predetermined IP address to connect to the system.
Once connectivity has been established, the administrator enters default credentials and accepts the license agreement. Once these general tasks have been completed, the real configuration work can begin. Hyper-converged systems almost always support manual configuration, but systems based around VMware's EVO:RAIL can automate the configuration process through the use of a JSON configuration file. EVO:RAIL systems can also use an out-of-the-box configuration in which the administrator provides the passwords.
If you are performing a manual configuration, you will need to configure a desktop or laptop computer to use an IP address that is on the same subnet as the default subnet used by your hyper-converged architecture. Using such an address allows you to connect to a hyper-converged server, but the configuration process is easier and more convenient if you have two workstations at your disposal.
The reason for this is simple: a hyper-converged architecture consists of multiple servers (often referred to as nodes). In many cases, the servers have to be brought online and configured one at a time. Part of this process involves assigning an IP address that fits into one of your network's subnets. When you change a server IP address this way, you lose connectivity to the server.
If you were working from a single configuration workstation, you would have to change the IP address, reconnect to the server and complete the configuration process. Once that is finished, you would have to once again change the workstation's IP address to bring the next node online. Using two separate workstations (or a workstation with two network adapters connected to two different subnets) prevents you from repeatedly reconfiguring the workstation's IP address.
In addition to configuring each node's IP address settings, you have to make some decisions regarding how the top-of-rack switch and available network adapters will be used. Each vendor makes its own recommendations, but it is relatively common for administrators to reserve one physical network adapter on each node for out-of-band management. Some vendors also recommend using VLANs as a way of isolating management traffic.
Once you have provisioned each node with basic network information, you need to provide other configuration information, such as:
- The name of the Windows domain
- The credentials to join that domain
- A host name
- The name of the failover cluster the server will participate in
In the case of an EVO:RAIL system, you need to provide a range of IP addresses that can be used as a virtual machine address pool. You may also have to specify a range of IP addresses to be used by vMotion and the virtual SAN.
Hyper-converged systems can vary widely in terms of the procedure used to set up storage. Each server typically includes its own direct-attached storage, and it is not uncommon for a vendor to automate the storage provisioning process. In the case of EVO:RAIL systems, however, storage is added to a VMware virtual SAN. This virtual SAN storage is automatically added to vCenter.
Hyper-converged trend has renewed the interest in local storage
Sifting through the hyper-converged hype
Hyper-convergence product roll-out requires updated job skills