Best Practices for Deploying Hyperconverged Infrastructure
These steps can help you overcome the challenge of migrating your IT infrastructure to HCI.
Many data center operators are interested in hyperconverged infrastructure, which combines storage, computing and networking into a single system. But making a major change to an organization’s IT infrastructure can be a daunting challenge.
It’s natural for IT managers and staffers to feel some trepidation at the prospect of fundamentally changing their IT architecture — even if only a portion of the enterprise’s workloads are going to run on the new solution, and even when the infrastructure is specifically meant to simplify operations. But fear of change shouldn’t keep organizations from moving forward. Most data center teams will find that they’re up to speed after just a little experience, training and research. As they get started, they should focus on these key considerations.
The industry site Hyperconverged.org suggests that data center administrators follow these four best practices for securing hyperconverged architecture:
- Guard against insider threats: Organizations should follow the “principle of least privilege,” granting the least possible level of access to individuals while still allowing them to do their jobs.
- Protect individual components: Although hyperconvergence integrates computing, storage and networking into a single cluster, Hyperconverged.org notes that criminals are still able to attack each component separately and recommends that organizations secure each individually. To this end, some vendors offer software-defined encryption that secures data both at rest and in transit. Also, fabric protection and shields for virtual machines can add security layers to the virtualization components of an HCI cluster. And backup software can be used for point-in-time infrastructure restores.
- Implement centralized security: Traditional data center security tools, which rely on full clients installed on each endpoint, are “too cumbersome” for HCI, Hyperconverged.org writes. Instead, the site recommends centralizing security through an agentless approach.
- Practice defense-in-depth: Finally, organizations should apply several layers of security to their infrastructure, protecting both the hardware and software from internal and external threats.
Preparation and Migration
Migrating to hyperconverged infrastructure involves moving virtual machines from one platform to another, and preparing these VMs for the shift is often the largest task tied to an HCI implementation. Administrators should identify any existing VM snapshots, determine which of their powered-off VMs to keep and scan data stores for any VMs and disk files that aren’t registered in their management tools. Each VM will need two migrations (one for storage and another for computing), which can be completed with specialized tools offered by IT vendors.
Preparing the physical environment is typically a less daunting task, but data center administrators still need to make sure that there is ample space for the new HCI nodes, as well as sufficient power and cooling for the infrastructure. Because hyperconverged infrastructure typically has a higher physical density than traditional platforms, HCI may generate the same amount of heat in a smaller space, requiring additional cooling.
Workload Placement and Management
While data center professionals who have never worked with hyperconverged infrastructure may be wary of the new architecture, the truth is that managing workloads for HCI doesn’t require many additional skills beyond what most IT staffers are likely to already possess. If a data center already incorporates any significant amount of server virtualization — and if staffers are able to manage that current environment — they’ll likely have little trouble managing hyperconverged infrastructure.
Typically, organizations find initial HCI workload placement more challenging than ongoing management. Data center administrators who lack experience with hyperconverged infrastructure may be unsure about which workloads are the best fit for hyperconvergence and may have questions about how to right-size a cluster for a given application. One use case that results in early success for many organizations is virtual desktop infrastructure. Because virtual desktops require more memory, computing and storage as they scale out, they’re a perfect fit for hyperconverged infrastructure, which incorporates all of these resources into a single, highly scalable solution.
Scaling Resources Independently
HCI nodes typically combine computing and storage resources, which raises the question: What if a cluster needs more computing power, or more storage, but not both? Adding a traditional HCI node results in “linear scaling,” with computing and storage both increasing in lockstep. However, not all workloads scale in a linear fashion. Many hyperconvergence vendors now offer compute-only and storage-only nodes, allowing data center administrators to scale out infrastructure in a more elastic manner.
Internal Expertise and External Assistance
Staff experience with virtualization is, perhaps, the most important expertise-related factor that will help ensure the success of a hyperconvergence deployment. However, many organizations have historically maintained siloed teams that separately manage networking, storage and servers. For obvious reasons, HCI is likely to significantly disrupt the organizational structure of such data centers, and staffers may need to be trained to work more effectively across these silos. Also, as hyperconvergence simplifies management, staffers may require additional training to prepare for roles that focus more on strategy and less on day-to-day infrastructure support.
A trusted external partner can help organizations that are new to hyperconvergence — both in designing and setting up a deployment and in training staff for their changing roles.
To learn more about how hyperconverged infrastructure can help your organization, read the CDW white paper “Hyperconvergence Goes Mainstream.”