Hyperconverged systems are designed to simplify server virtualization. These systems consist of matched computing, networking and storage hardware, as well as a software layer that is certified to work with the hardware. Rather than individually purchase and configure the necessary hardware and software components, IT professionals can purchase a hyperconverged system to simplify the implementation process while also ensuring an optimal configuration.
Even though hyperconverged systems are occasionally described as plug-and-play virtualization, or virtualization in a box, deploying a hyperconverged system usually involves quite a bit of work. Yes, hyperconverged systems are easier to deploy than a la carte server virtualization, but it’s unrealistic to expect to plug in an appliance and be up and running in five minutes.
Unfortunately, there is no standard process for deploying hyperconverged systems. Some tasks are consistent across the board, but there are also vendor-specific nuances. For example, Hyper-V systems may require administrators to choose between using CredSSP or Kerberos with constrained delegation as an authentication protocol for live migrations. Regardless of which vendor’s product you purchase, there are best practices to adhere to as you prepare to deploy a hyperconverged system.
1. Get Politics Out of the Way Up Front
One of the most important things that you can do as you prepare to deploy a hyperconverged system is to address the issue of ownership before you ever begin the configuration process. Larger organizations divide IT into technology silos with groups responsible for networking, storage, Active Directory, etc. Hyperconvergence tends to break this administrative model because a hyperconverged system is designed to be self-contained. It is important to meet with your organization’s various IT groups ahead of time to establish ownership boundaries. For example, because storage is integrated into the hyperconverged system, you need to determine if the networking team or the storage team will manage it.
2. Determine How Network Interfaces Will Be Configured
Perhaps the most important predeployment task for hyperconverged systems is to plan for network configuration. On the surface, it may seem the system can be connected to a top-of-rack switch and assigned a series of IP addresses, but there are actually a number of decisions that must be made. For example, EVO:RAIL systems require the administrator to make decisions such as:
- Whether to use one switch or two
- What host names to use
- How the management virtual LAN should be configured
- How to support out-of-band management
- What IP address should be used by vCenter
3. Compile a Deployment Checklist
A deployment checklist can be really valuable as you make decisions about the hyperconverged system configuration. Not only does it help you to document your configuration decisions, it also helps you to avoid missing steps. Some vendors provide a checklist that you can follow; for instance, VMware offers a downloadable checklist for its EVO:RAIL systems.
4. Test, Test, Test
As you configure your hyperconverged system, it is important to run tests to verify that everything is configured properly. The actual tests that you should run vary depending on which vendor’s solution you are using. At a minimum, you should test the network configuration (connectivity, Dynamic Host Configuration Protocol, Domain Name System, and so forth) and take advantage of any built-in diagnostic features. For example, Windows-based systems include a mechanism for performing a series of tests designed to validate the failover cluster configuration.