Yes Virginia, server hardware architecture can be disaggregated with HPE ProLiant Compute

[ComputeExperts]

HPE’s HPE ProLiant Open Compute Project Data Center Modular Hardware System (DC-MHS) delivers accelerated time-to-market, agility, and cost efficiencies to Service Providers. These new 12^th generation HPE ProLiant servers are optimized for various workloads, from enterprise and edge AI to large-scale Infrastructure-as-a-Service and are compliant with specifications of the Open Compute Project (OCP)[i]. The servers accommodate the shift from monolithic server architectures to modular, scalable, and pluggable designs, focusing on interoperability across the data center by providing consistent interfaces and form factors.

A new paradigm

Service Providers (SPs) have long understood and embraced cloud native architecture for their application development and deployments. Prior to this transition to a new disaggregated application architecture, SPs had to deal with application lifecycles that were cumbersome, rigid, and bloated for many reasons. These included:

• Applications written as one monolithic code base
• Redeployment of the whole application for even a small change, and
• Long and onerous development and testing cycles.

With the transition to disaggregated cloud native development, the application lifecycle became lighter, nimbler, and leaner – delivering multiple benefits including rapid innovation, scalable design for agile reconfiguration, and flexible, loosely coupled systems.

Figure 1 illustrates the progression from traditional to cloud native applications architecture.

Figure 1

This architectural approach has now been extended to computing in the form of disaggregated server hardware. The recently announced, and now ramping, HPE ProLiant Compute DL320/DL340 Gen12 SP servers are HPE’s first platforms that deliver this modular, disaggregated computing hardware.

Figure 2 illustrates this dramatic transformation from traditional monolithic server designs.

Figure 2

These new HPE ProLiant servers conform to the Data Center Modular Hardware System (DC-MHS) specification developed under the auspices of the Open Compute Project, an industry group, of which HPE is a founding member. In addition to HPE, the group includes industry stalwarts such as Intel, AMD, NVIDIA, Microsoft, Google, Meta and Dell.

Embracing modular server design

Let’s take a deeper dive into this exciting and embryonic transition to modular server design.

Think of a traditional server being a “monolithic” or “planar” design, having the following characteristics:

• All components are statically defined on the motherboard.
• There is an inability to evolve individual modules (subsystems) independently without whole redesign.
• Server generation is tied to a specific processor family.

In contrast, a DC-MHS compliant server displays the characteristics like those of cloud native applications, such as:

• The motherboard is deconstructed into individual modules.
• Individual subsystems that can evolve independently.
• Server lifecycle that is untethered from the lifecycle of any individual module.

Consider, if you will, that in addition to modularity, security, and interoperability, other critical success factors for broad adoption are being facilitated by members of the Open Compute Project – even as you read this blog! HPE is a core team member of both the Data Center Secure Control Module (DC-SCM), and the Data Center Plug and Play Module (DC-PnP) workstreams.

DC-SCM is the main point of control for DC-MHS ecosystem. DC-SCM covers management, control, AND the root of security for the system. It includes DC-SCM System management Baseband Management Controller (BMC) on a standardized modular plug-in board. The DC-PnP workstream, led by HPE, will drive the OCP initiative to eventually deliver what could be visualized as the “Nirvana” of interoperability: the ability to swap a subsystem from Vendor A, with one from Vendor B.

HPE is proud to support OpenBMC on HPE ProLiant servers, enabling customers to securely customize their BMC for their specific requirements. Readers interested in learning about DC-MHS and other workstreams not discussed here, are encouraged to visit the OCP website.

A 12^th generation first

To recap, the new HPE ProLiant Compute DL320 Gen12 SP and DL340 Gen12 SP servers are designed to give Service Providers greater architecture flexibility by right-sizing infrastructure with seamless deployment for long-term investments. These platforms deliver multiple benefits to Service Providers including:

• Operational agility. Reusable building blocks across multiple designs for multiple markets
• Accelerated time to market. Minimizes the need for ground-up custom designs, and
• Lower Total cost of Ownership. Module innovations, sourcing, and support are shared across platforms.

We invite you to learn more about these servers here and view a short video chalk talk.

Meet our authors, J.B. Knight and Rahul Shah

JBK has been part of the extended HPE family since the late ‘90s. His career began we he was hired as a Digital UNIX Systems Engineer with Wyle Systems, a VAR of Digital Equipment Corp (DEC). He joined Compaq in 2001 covering Tru64 UNIX and AlphaServers as a regional Presales resource. 

With the HP/Compaq merger, JBK morphed into a Mission-Critical Solutions Architect, working for an extended period as a Proof-Of-Concept Specialist for HP-UX, Serviceguard, and Oracle RDBMS. In 2005, he pivoted to a ProLiant Presales position and covered Enterprise accounts in Southern California. Since 2019, he has served as both a Compute Go-To-Market Specialist covering North America, and now as an Engineer and Business Program Manager for T2/T3 Service Providers.

Rahul Shah is currently a product marketing manager at Hewlett Packard Enterprise (HPE), focusing HPE ProLiant servers. Working in the data center IT industry since 2008, Rahul has extensive product management and product marketing experience with enterprise technology companies including QLogic, Emulex, Broadcom, and Pure Storage. Rahul’s education includes an MBA from the Kenan Flagler business school at the University of North Carolina, Chapel Hill as well as a BS in Electrical Engineering from the MS University, Vadodara, India.

Compute Experts
Hewlett Packard Enterprise

twitter.com/hpe_compute
linkedin.com/showcase/hpe-servers-and-systems/
hpe.com/servers

[i] The OCP Foundation is responsible for fostering, serving and seeding the OCP Community to develop new open solutions that can meet the market and shape the future. In shaping the future, OCP will continue to invest in strategic initiatives that prepare the IT ecosystem for major changes, such as AI & ML, optics, sustainable data center solutions, advanced power management and cooling techniques, composable silicon and sustainability. https://www.opencompute.org/