Is hyperconverged infrastructure right for your business?
As delivery solutions architects, many engineers get into deep conversations with customers regarding their technology roadmap and overall business initiatives while performing delivery engagements onsite. One popular topic that often arises is hyperconverged infrastructure (HCI).
Many organizations are intrigued by HCI, while others are a bit hesitant, instead choosing to trust their proven three-tiered architecture due to various business and workload requirements. The questions our solution architects field most frequently are:
- Is HCI a radical idea that won’t pan out long-term?
- Will HCI support all our mission-critical workloads and hybrid cloud initiatives?
- How will HCI impact my budget?
- How does HCI perform, and does my staff need to be trained?
- What are the concerns and pain points of an HCI, and how do I avoid them?
The best way to approach these questions is to look at how hyperconverged infrastructure differs from traditional three-tiered architecture.
Hyperconverged vs. three-tiered architecture
A hyperconverged system combines compute, storage, and network with several hypervisor options. The “nodes” share their local storage over a private network, forming an available cluster without the investment of shared storage.
With three-tiered architecture, compute, shared storage, and networking are purchased separately to create the same cluster.
The simplicity and cost savings of the HCI model are undeniable. Since HCI is software-defined and easy to install, it reduces both capital and operational expenses while delivering comparable performance and availability to three-tiered architecture.
With three-tiered architecture, the cost of building and maintaining the compute and storage infrastructure is higher for both capex and opex expenditures.
Due to data locality, hyperconverged systems compete with even the fastest storage area network (SAN).
While many times larger, a shared storage array must deal with the latency of the fabric infrastructure to send and receive data (north-south traffic). A hyperconverged host can obtain the same data locally (east-west traffic), without traversing a fabric network or additional network hops, which triggers latency.
HCI is not only here to stay, but it also has a place in every organization’s data center.
But is hyperconverged the future for every situation? That’s more complicated.
Technology changes every day, so what is true today might not apply tomorrow. There are both benefits and drawbacks to HCI. Before making decisions, it’s important you understand both and how HCI relates to your business requirements and overall initiatives.
The benefits of hyperconverged infrastructure
Let’s start with the positives. There are three main benefits to HCI:
- HCI solutions provide excellent management and monitoring tools in a single box, allowing for simpler troubleshooting, monitoring, maintenance, upgrades, and provisioning. The entire stack is software defined. Automation tools have far fewer objects and components to interact with. A single systems administrator can support the entire stack, and compatibility is not an issue.
With HCI, one vendor supports everything. There’s no passing the buck from vendor to vendor when support is needed. Lifecycle management usually includes hardware firmware updates, as well as software updates. All combinations of hardware, firmware, and software are tested, validated, and supported.
- Cost savings. By purchasing fewer components and managing them from a single pane of glass, HCI reduces capex and opex. There are no additional procurement or support costs for storage arrays and fibre channel (FC) switches. There’s also reduced need for:
- Rack space, resulting in lower data center costs
- Network switch ports, resulting in fewer switches (and their associated maintenance costs) overall
- Power connections, resulting in fewer power distribution unites (PDUs) and (uninterruptible power supply (UPS), leading to power and cooling savings
In addition, there is much less planned downtime for upgrades and maintenance, which allows the organization to perform these without much disruption to the business, while keeping applications online. Moreover, with HCI’s single pane of glass management, there is much less need to have a dedicated SME and extensive training to administrate the environment, thus shifting the good portion of administration costs toward more business-related projects and tasks, as defined by the enterprise.
- Built-in cloud agility. Seamless cloud integration is a core component of HCI. Moving workloads with a single click of a button to and from the public cloud allows for unprecedented scalability and flexibility. While this is certainly possible with three-tier technology, many HCI vendors license this type of software in their standard package offering, right out of the box.
The drawbacks of hyperconverged infrastructure
We discussed what HCI solutions do well. Now, let’s look at the drawbacks. There are a few worth noting:
- Scaling granularity. While HCI offers both compute and storage-centric nodes, you must always purchase a small amount of compute resources as you scale linearly, usually by adding a new node. The cost could be greater than adding disks to a storage array. Scaling granularity and flexibility is the keystone of a three-tiered solution.
- Vendor lock-in. While a single vendor may be considered an HCI advantage, the multi-vendor approach has added advantages as well. Other vendors may offer attractive features and pricing. If you are locked in to a single HCI solution, it may be difficult to leverage other technologies and vendors. Make sure you’re fully aware of the roadmap and upgrade rules when choosing an HCI solution.
- External HW connectivity. These days, nearly every server can be virtualized, but there are still instances where physical servers and clusters are required. In these situations, while some hyperconverged solutions can present block storage via iSCSI, others may create silos in your infrastructure, where a single SAN array can service both a physical and virtual environment with ease. Also, not all hyperconverged systems support connectivity to existing storage arrays.
Is hyperconverged right for your business?
There are a lot of factors to consider.
If simplicity and cost savings are your top priorities, HCI might be your best bet, as it can significantly reduce your IT footprint and save capex and opex if done correctly. On the flip side, if you must remain flexible in growth and technology while serving many different workloads that utilize both physical and virtual servers, three-tiered options might be a better approach. You may even find that the optimal approach is to leverage both technologies.
Hyperconverged is no longer a disruptive idea; it’s an essential component of a complex and evolving IT landscape. HCI and three-tiered architecture can co-exist while providing a robust, hybrid cloud environment to effectively meet every business requirement. There are many vendors that offer their hyperconverged solutions, including Nutanix, Dell EMC VxRail, VMware vSAN ReadyNodes, HPE SimpliVity, Cisco HyperFlex, Azure Stack HCI, Scale Computing, Pivot3 and a few others . HCI solutions are designed for a wide range of workloads and may not be the best solution for specific use cases.
Before choosing a solution that best suits your business needs, make sure you evaluate your organization’s long-term strategy and business requirements, including growth predictions, use of public cloud, power, cooling, skills/training, and more.
SHI’s technical expertise can help you optimize your solution based on your business requirements and goals. Contact us to start a conversation.