Converged and Hyperconverged Infrastructure – Comparisons and What is Best for Your Organization?
ParkView Managed Services
IT professionals are under intense pressure to eliminate inefficiency and waste in their IT infrastructure. However, as infrastructures have evolved, this pressure has been alleviated by technology vendors devising new solutions that enable data centers to perform better while cutting down on space and energy use. With the hardware replacement cycle coming up typically every three- five years, enterprises have regular opportunities to replace older equipment and architectures with something better.
Exampled of this are the traditional tiered approaches to infrastructure, which keep compute and storage in separate silos, instead giving way to converged infrastructure (CI), and increasingly, hyperconverged infrastructure (HCI).
CI and HCI are certainly catching on. Industry research reveals that the market for converged infrastructures is on track to grow at 24% year over year, going from $9 billion in 2021 to a projected $52 billion by 2029. HCI, now a $13.37 billion market, is forecast to grow at 25% per year and pass the $40 billion mark by 2030. Both sectors are growing far more quickly than the overall enterprise IT sector, which has a growth rate of around 9% per year, according to Gartner.
Below we explore convergence vs hyperconvergence. Specifically, we discuss how these two technologies work, the converged and hyperconverged infrastructure difference, and why it may be a good idea to deploy them in certain workloads.
Before this however, we must understand the baseline alternative – the traditional infrastructure, otherwise known as a non-converged infrastructure.
What is Traditional Infrastructure?
What is a Converged Infrastructure?
5 Benefits of Converged Infrastructures
6 Challenges of Converged Infrastructures
What is a Hyperconverged Infrastructure?
5 Benefits of Hyperconverged Infrastructures
5 Challenges of Hyperconverged Infrastructures
Evolution from Traditional Infrastructures
What is the Difference Between Converged and Hyperconverged Infrastructure?
When to use a Converged vs Hyperconverged Infrastructure?
How to Implement Converged Infrastructure and Hyperconverged Infrastructure
Converged Infrastructure Implementation
Hyperconverged Infrastructure Implementation
Undecided with Converged Infrastructure vs Hyperconverged Infrastructure?
What is Traditional Infrastructure?
A traditional IT infrastructure comprises servers (compute) and storage in separate tiers, each running on its own dedicated hardware. Network switches, also on separate hardware, connect compute and storage elements. Separate management tools control the operations of the compute, storage, and network tiers of the infrastructure, as each must be managed and scaled out separately.
There is nothing wrong with a traditional infrastructure, yet it has drawbacks. It tends to be complex to set up, manage, and scale. There may be costly and cumbersome storage solutions, such as storage area networks (SANs) and network attached storage (NAS), which require dedicated maintenance and scaling efforts as workloads evolve.
When assessing converged infrastructure vs traditional, or even hyperconverged infrastructure vs traditional, traditional’s primary disadvantage is that it is considered an outdated setup compared to CI and HCI.
What is a Converged Infrastructure?
Converged infrastructure, converges — or merges — compute and storage into a single piece of hardware. Instead of separate racks for servers, the storage network, and storage systems, CI has all three functions in a single appliance.
A converged infrastructure server/storage appliance has a computer server and a certain amount of onboard storage, e.g., an X-86 server with 8 Gigabytes of RAM and a 1.92 terabyte solid state drive (SSD) all on one chassis. Infrastructure managers can buy as many appliances as they need and manage them through a unified compute/storage management console.
5 Benefits of Converged Infrastructures
A Converged infrastructure offers infrastructure managers a host of benefits, most notably its “plug and play” nature. Appliances can be used individually or joined in dynamic blocks like Lego pieces. As a result, enterprises that deploy CI experience:
- Fewer silos – There are no more separate processes for procurement, installation, maintenance, and admin for compute, storage and network—each with its own vendors, skillsets, and team members.
- Lower hardware support and maintenance costs – Converged infrastructures typically lead to infrastructure vendor consolidation and the procurement of less heterogeneous systems. This is all easier to support, usually requiring fewer people and outside resources. Fewer vendor compatibility issues also contribute to lower support and maintenance costs.
- Better agility – Changing or scaling traditional infrastructure can be a slow and costly process. In contrast, CI’s plug-and-play capability makes it relatively simple and quick to modify or expand. It becomes easier and faster to deploy IT infrastructure in support of agile business moves.
- Procurement and deployment become faster and simpler – CI’s all-in-one-box essence makes it comparatively streamlined versus traditional infrastructure, whose procurement requires a separate process for specifications, vendor selection, and acquisition, followed by an equally siloed deployment process.
- Growth on demand – A converged infrastructure makes it possible for enterprises to design, build, and scale out pieces of a virtualization stack without affecting other areas of the infrastructure. In contrast, with traditional infrastructure, a change in one place, like the SAN, can set off a round of changes elsewhere. Such dependencies slow down processes of growth and change.
6 Challenges of Converged Infrastructures
A converged infrastructure is not a flawless panacea. It is not ideal for every job, and the technology comes with its share of challenges. For instance:
- Getting locked in with a vendor — Implementing converged infrastructures may lead to reliance on a single vendor for infrastructure, a lock-in situation that can lead to friction, contractual issues, and so forth.
- Dealing with platform limitations — Not every workload goes with CI, for example, a high-performance computing (HCI) processes may be better suited on purpose-built infrastructure.
- Misalignment of storage capabilities and capacities — Some workloads require storage at performance levels and capacities that don’t match the onboard storage in CI appliances. This can lead to inefficiencies, performance problems, and unexpected costs.
- Handling more than one vendor — Even if the appliance is converged, the storage and compute elements may come from different sources, e.g., the SSD maker and CPU maker will not be the same company, and warrantee, upgrade, and support differences can emerge as the CI is deployed.
- Requiring specialized skillsets — a converged infrastructure is different from traditional infrastructure, so people will have to master new skills to manage it. This may not be a big issue, but it will create a change management process to oversee.
- Running into higher costs — CI appliances may cost more, in terms of compute and storage capabilities, than traditional infrastructure solutions. However, the overall costs of running the CI may be lower due to new efficiencies.
What is a Hyperconverged Infrastructure?
A hyperconverged infrastructure, sometimes referred to as ultra-converged infrastructure, is a software-defined approach to infrastructure. In contrast to converged infrastructures, HCI uses software to set up infrastructure instances from pools of compute and storage resources.
Hyperconverged systems can be deployed on any kind of hardware, including “commodity” equipment, along with any type of hypervisor. HCI also works through APIs, which enable dynamic management and orchestration; and its management software can translate virtual machines (VMs) between hypervisors and across platforms. Hyper-converged infrastructures can thus integrate cloud and on-premises environments.
5 Benefits of Hyperconverged Infrastructures
The benefits of hyperconverged infrastructures arise mostly from the decoupling of its hardware elements from its software management layer. Highlights include:
- Greater IT operations productivity — With a single control interface and no hardware to procure and deploy, HCI allows for moving quickly with IT projects, including putting new applications into production.
- Lower costs — With fewer interoperation issues and streamlined management, HCI is often inexpensive to operate. This is a key consideration in the hyperconverged infrastructure vs converged discussion.
- Scalability — Hyperconverged architectures can scale quickly and easily because all that is required to expand a system’s footprint is the software-driven allocation of compute and/or storage from the resource pool.
- Efficient management — With a single, centralized software control interface, HCI makes infrastructure management highly efficient.
- Potential to avoid third-party replication solutions — Hyperconverged architectures enable built-in replication between HCI nodes. It is not necessary to deploy third-party replication solutions, though there may be reasons for doing so.
5 Challenges of Hyperconverged Infrastructures
Hyperconverged infrastructures have different challenges to those of converged. For instance:
- Cost impact — HCI may raise certain costs, such as when software licensing expenses pile up with the use of more CPUs than would be needed in a traditional setup.
- Scalability — HCI is designed for scalability, but some workloads are not a good fit with the scale-out mode of expansion. For example, certain compute-intensive applications need to scale up, so they’re a poor fit for a hyperconverged infrastructure.
- Unnecessary hardware — HCI vendors may require the purchase of compute and networking hardware even if the need is only for more storage.
- High power usage — Hyperconverged systems can create power load problems in data centers due to hardware density.
- Data center redundancy — Don’t assume redundancy is built into HCI environments. Such a capability has to be specifically architected and implemented.
Evolution from Traditional Infrastructures
Converged and hyperconverged infrastructures represent an evolution from traditional infrastructures, specifically hyperconverged infrastructure vs traditional, is one of today’s common considerations of analysis in data center design.
Converged infrastructure and hyperconverged infrastructure take infrastructure away from the old practice of separating of compute, storage, and networking into separate equipment stacks and management silos. The results are better efficiency and greater agility. However, it would be a mistake to take the “newer is always better” view of converged/hyperconverged infrastructure.
Indeed, many workloads are more suited to work on traditional infrastructure. An enterprise resource planning (ERP) solution with a massive in-memory database, for example, is not a good candidate for converged infrastructures or hyperconverged infrastructures. Nor is a storage-heavy, low compute workload like video archiving. In that case, it is far better to build a bespoke infrastructure that leverages low-power ARM processors and lower-performing storage for optimal cost and power characteristics.
What is the Difference Between Converged and Hyperconverged Infrastructure?
Converged and hyperconverged infrastructure have certain things in common, but they differ in many ways. The following table offers a high-level view of the difference between converged and hyperconverged infrastructure.
Converged Infrastructure (CI) | Hyperconverged Infrastructure (HCI) | |
Definition | An approach to infrastructure that uses appliances that combine compute (servers), storage, and networking. Components may come from more than one vendor | A software-driven approach to infrastructure that provides a centralized software point of control for converged pools of compute and storage. HCI can work using commodity hardware. |
Architecture | Converged architecture is modular, i.e., admins can manage or scale each component independently. | A software-defined architecture features tight integration of the components, each of which represent a node in the HCI architecture. |
Scalability | Scale-up, wherein it is possible, to add more appliances (i.e., modules) to expand a system. | Scale-out, with growth achieved by adding nodes to the HCI cluster. |
Management | An overall converged infrastructure management solution enables control over the converged infrastructure environment, but each component (e.g., storage) may have its own independent management toolset. | A single, unified management interface enables centralized management of the hyperconverged environment. |
Cost | Initial cost may be higher than traditional alternatives, but CI can be economical for large environments. | HCI costs can be lower than CI due to use of commodity hardware. Scale-out costs tend to be predictable. |
Flexibility | Infrastructure managers can select components from multiple vendors and use a mix-and-match approach, which enables a high degree of flexibility. | HCI’s tendency for hardware standardization can limit customization and flexibility. The software-defined nature of HCI, however, enables flexibility not available in alternatives. |
Software-Defined Storage | It is possible, depending on the vendors, but not automatically included. | Possible and relatively simple to deploy given the software-defined nature of the infrastructure. |
When to use a Converged vs Hyperconverged Infrastructure
Assessing when to utilize a converged/hyperconverged infrastructure can be difficult because in many ways they are similar, yet distinct from one another in that each is appropriate for different use cases.
There is some overlap between converged infrastructure vs hyperconverged, as both technologies are good for virtual desktop infrastructure (VDI), server consolidation, and remote office/branch office (ROBO). However, there are clear instances where one infrastructure setup is better suited than another, when evaluating a converged infrastructure vs hyperconverged infrastructure.
When to Use a Converged Infrastructure
A converged Infrastructure makes sense for use cases such as:
- Remote disaster recovery (DR) sites — DR sites should ideally be easy to deploy and manage. Complexity and management challenges make them less reliable, which goes against their purpose. For these reasons, converged infrastructures can be a great choice for a remote DR site.
- Data center consolidation — Data centers often devote excessive rack space to under-utilized hardware, which leads to a waste of power and cooling, as well as sub-optimal data center financial outcomes. CI appliances are a great way to combine under-utilized solutions into a more compact form factor.
- Application virtualization — The old mode of running a single application on a dedicated server almost always leads to highly inefficient use of data center resources. Virtualizing applications solves this problem, and a converged infrastructure is an effective way to achieve this goal.
When to Use a Hyperconverged Infrastructure
On the other hand, compelling hyperconverged infrastructure use cases include:
- Edge computing — Edge instances sometimes need space-efficient storage that can support multiple servers. HCI is a good fit for this use case because edge data centers are often very space- and power-constrained, for example, 120 KW or smaller. Hyperconverged infrastructures help by enabling edge managers to pack a lot of systemic capability into a small space.
- Test and development — HCI enables dev and test teams to set up small-scale environments quickly, which saves money and time in processes that demand economy and speed.
- Logging and analytics — These processes generate a large amount of data, with constant growth in data storage requirements. The scale out nature of hyperconverged infrastructures is a good fit as a result.
How to Implement Converged Infrastructure and Hyperconverged Infrastructure
Converged infrastructure and hyperconverged infrastructure implementation have a lot in common. There’s the process of determining which workloads are suitable for each approach to infrastructure; and there’s creating a reference architecture and spec’ing out the compute, storage, and network parameters required. Both also need detailed analysis of disk input/output (I/O) capacities.
Establishing a business case is a good idea, so there is executive buy-in and more than a little thought to future directions and how the proposed infrastructure will serve probable needs as time goes on. The converged and hyperconverged infrastructure difference in implementation arises due to differences in the design and functioning of the respective approaches to infrastructure.
Below we assess the different implementation steps for a hyperconverged infrastructure vs converged.
Converged Infrastructure Implementation
Assuming the reference architecture and requirements are ready, converged infrastructure implementation involves first selecting the appropriate vendor, or vendors, and then the right specific appliance.
It’s a good idea to deploy a small version of the proposed CI first and assess how well it works. This is a good opportunity to work out unexpected kinks or address skills shortfalls. Phased deployment should follow.
One of the great things about having your Converged Infrastructure solution installed is that it will come pre-rack-and-stacked by the vendor, with no intra-rack cabling needed. Once the rack is moved into its desired location in the data center, all IT teams must do to complete the install is choose from a selection of pre-validated configuration scripts — depending on the desired workloads — and then carry out some basic networking.
Once up and running, the onus is still on the vendor to help keep all software and hardware (OS versions, microcode, firmware, patching) up to date and in sync for the compute, networking, and storage inside.
Hyperconverged Infrastructure Implementation
Hyperconverged infrastructure implementation is about spec’ing and configuring compute and storage resource pools, where pools must match workloads. A phased implementation is also recommended, though that can occur after the full-scale pools have been deployed.
With HCI, there’s no external storage arrays or no external SANs to worry about; only servers and switches, for high availability. The installation required for HCI is not much different than adding traditional servers into your existing environment.
Servers are racked and cabled into the existing HCI switch, and joined to the cluster. Additionally, with the continued drop in price of NVMe SSDs, there’s not much to tune storage performance-wise either, just creating one large pool for all applications to play in.
Undecided with Converged Infrastructure vs Hyperconverged Infrastructure? Get a Partner that Understands Both
When evaluating convergence vs hyperconvergence, or even considering if a non-converged infrastructure is right for you, there is a lot to consider. Park Place Technologies can support you no matter what, regardless of your infrastructure set up.
Our data center maintenance service can support your non-converged/traditional infrastructure, by keeping your devices maintained at a lower cost compared to the OEM, with no compromise in support quality. If you are looking to invest in a converged infrastructure vs traditional arrangement, perhaps as part of disaster recovery solution, our Professional Services can help, with an expert team on-hand to deliver IT assistance to infrastructure teams.
Going one step further and managing a virtualized HCI infrastructure? Infrastructure Managed Services from Park Place Technologies ensures your operating system software and virtual platforms operate at peak efficiency. Your HCI cluster will benefit from our incident management, patch management and remediation service — guaranteeing your entire hyperconverged infrastructure is supported.
Contact Park Place Technologies today to learn how we can support and maintain your traditional, converged and hyperconverged infrastructure.