Data Center Strategy: Composability vs. Converged & Hyperconverged Infrastructure

Posted on December 12, 2020 by rawee.k

Based on continually changing IT and business needs, the challenge to meet the growing number of on-prem applications and services – including artificial intelligence, dynamic cloud, IoT and edge computing – has forced organizations, like  to rethink the traditional static architecture data center.

Most recently, a data center administrator’s speed and flexibility needs may have been met through a converged or hyperconverged infrastructure (HCI). However, both have scalability limitations: When you need more compute, storage or networking, you have to add another HCI server node. With composable infrastructure, you simply add resources when you need them. As a result, composable infrastructure adoption by data-driven organizations has grown in an effort to unlock the full value of its technology investment.

Previously, we covered what is composable infrastructure and how organizations can benefit from composability. Next, we’ll cover how data centers built with software-defined composable infrastructure architecture compare to ones based on converged and hyperconverged infrastructures (HCI).

Choosing between software-defined architectures? Here’s what data center administrators need to know to compare composable infrastructure versus converged and hyperconverged infrastructure. Questions? Drop us an email at sales@pogolinux.com

Converged Infrastructure Overview

In a converged infrastructure environments, a pre-configured package of software and hardware are housed in a single, integrated software-defined architecture that has been optimized for hosted application performance, availability, and scalability. In addition to a simplified procurement process, converged solutions are easier to deploy, administer and manage than a traditional client-server approach with rackmount servers, network storage and network devices attached.

Data center users have typically deployed convergence to address a specific data center application or workload, such as server virtualization, databases or line-of-business applications. For virtualization, although converged infrastructure delivers performance gains due to compute, storage and networking being physically integrated, the management of those discrete resources often remains silo’d.

Convergence Management Challenges

While converged solutions help organizations bypass the integration and configuration steps required to deploy virtualized infrastructure, they often require administrators to use multiple interfaces to manage each converged resource. In addition, these offerings typically come at a premium price from name brand suppliers.

Hyperconverged Infrastructure (HCI) Overview

Hyperconverged infrastructure (HCI) solutions are “purpose-built” to implement an integrated virtualized infrastructure. Software-defined elements are implemented virtually, with seamless integration to the hypervisor with the ability to move a workload from one host to another.

HCI systems are pre-assembled within a single enclosure with hardware and virtualization software with localized compute, clustering, networking, and direct attached storage (DAS) to implement data resiliency. This makes management and administration a lot simpler than converged systems, which consist of multiple components cabled together a server rack.

The Scalability Limitations of Hyperconvergence

While effective for specific workloads, hyperconverged doesn’t solve all a data center operators need in the long-run. For instance, typical HCI environments only scale out up to 30 nodes, which means it may not scale out to the level that . As a result, IT department’s need to deploy additional modules to keep pace with the uneven demands of virtualized environments.

The benefits of HCI for virtualization becomes significantly diminished when the system performance becomes uneven. Without the ability to disaggregate resources through software API’s, the HCI solution is unable to dynamically adapt to changes in unpredictable, virtualized workloads. As a result, some high-value resources sit idle while others are taxed to their limits.

Composable Infrastructure Overview

The fundamental benefit of composability is to give IT departments the ability to maximize resource utilization in the data center. Also called “disaggregated infrastructure”, composable infrastructure takes the concept of HCI a step further with more fluid approach to hardware management and provisioning.

Composable infrastructure is the ability to connect compute, GPU, storage and networking into pools of resources that are provisioned through software APIs and network fabric, then deployed into bare-metal server nodes. In composable infrastructure environments, hardware resources can be composed into exact-sized system designs (including bare-metal) for each private cloud hosted application’s workload needs. This infrastructure architecture approach is similar to a public cloud where resource capacity is requested and provisioned from shared capacity – except composable infrastructure sits behind a corporate firewall on-premises in an organization’s data center.

Composable infrastructure is the ability to connect compute, GPU, storage and networking into pools of resources that are provisioned through software APIs and network fabric, then deployed into bare-metal server nodes. In composable infrastructure environments, hardware resources can be composed into exact-sized system designs (including bare-metal) for each private cloud hosted application’s workload needs. This infrastructure architecture approach is similar to a public cloud where resource capacity is requested and provisioned from shared capacity – except composable infrastructure sits behind a corporate firewall on-premises in an organization’s data center.

The Difference is Dynamic Resource Sharing  & Flexible Provisioning

Composable IT offers the same integration and automation capabilities as HCI while offering greater levels of abstraction. Data center operators can dynamically share and rapidly provision server resources with the agility and flexibility of a public cloud via software-defined management API’s, network switches and HBA cards.

Like an HCI solution architecture, composable IT combines compute, storage and networking components into a single platform, but it’s not preconfigured to address a specific workload. Physical assets are logically pooled, so data center administrators can mix and match disaggregated components and connected devices from their commodity server infrastructure to support a software’s specific hardware requirements.

Compose a complex architecture of bare-metal server environments on demand across PCIe, Ethernet, and InfiniBand network fabrics without manually-configuring resources. To scale out a composable infrastructure, simply add individual components – including CPU’s, NVMe SSD’s, GPU’s, NIC’s and FPGA – which is a cost-effective departure from adding an entire HCI server node.

Choosing Between Software-Defined Architectures

Hosted applications drive the data center market and factor heavily into any purchasing decisions. Current and future applications, workloads and specific use cases must be considered along with performance and capacity requirements. In many scenarios, these technologies are complementary and together provide the right mix of features and flexibility.

For data center infrastructure and operations leaders, composability trades away the convenience of a purpose-built infrastructure offered by converged or hyperconverged, in exchange for the ability to deliver on-premises applications and services with the same speed and agility a public cloud.

Ultimately, a thorough understanding of the data center applications being run is required because it comes down to each organization’s data center needs. More than likely, the decision to deploy a composable infrastructure vs. a converged or hyperconverged infrastructure is not necessarily an either-or situation, as the answer will depends greatly on what data center operators are trying to accomplish.

Choosing the Right Composable Infrastructure Solution

The ability to accelerate time-to-market for digital products, or improve service-levels and project delivery times, is a competitive advantage that’s necessary for any data-driven organization. A comprehensive software-defined composable infrastructure framework can providers data center admins the ability to configure, manage and scale out physical bare-metal server systems in seconds.

Setting up a Composability Proof of Concept

Hosted applications will drive future adoption of composable infrastructure, and proof of concepts (POC’s) are the best way to evaluate a composability investment against a converged or hyperconverged environment. More often than not, the right solution for your data center is a mix of composability alongside a static architecture to experience the greatest platform flexibility, scalability, and unit economics.

IT decision makers should lay out rigorous performance, resiliency and scalability requirements. Additionally, knowledge of the hosted application requirements is necessary to ensure a POCs provide clear, decisive results. To objectively measure technology and operational benefits, it’s possible to start small and build out as you go. Most vendors composable infrastructure will happily help you set up a proof-of-concept solution to test out the design set-up that best meets your needs.

If you’d like to learn more about how to maximize your data center infrastructure by up to 90% while minimizing its footprint, give us a call at (888) 828-7646, email us at sales@pogolinux.com or book a time calendar to speak. We’ve helped organizations of all sizes deploy composable solutions for just about every IT budget.

composable infrastructure