In their efforts to adapt to the demands of the digital economy, the Internet of Things, and other disruptive changes, datacenters are facing big technical challenges in terms of flexibility and scale. This is all because of traditional rigid architectures.

Today’s hardware infrastructure for datacenters typically comes as preconfigured 1U or 2U servers with individual processors, memory, I/O, and network interface controller (NIC). To upgrade or add to this infrastructure, a complete system needs to be built and integrated into the rack, and connected via management and virtual pooling. This system will essentially operate as a single unit of compute, meaning its internal resources of CPU, memory, and dedicated storage are accessed solely by that server, locking down resources that are not always fully utilized.

To complicate the challenges, the conventional server architecture is in general a vertical deployment model, with many different hardware/software models present for management. So how can you overcome rigid, expensive, time-consuming datacenter build-outs that can’t keep pace with the digital demands of today? The answers are already here—in the form of disaggregation of the datacenter rack.

With this new approach to the rack, a logical architecture disaggregates and pools compute, storage, and network resources and provides a means to create a shared and automated rack architecture that enables higher performance, lower cost, and rapid deployment of services. At this point, agility at hyperscale is no longer a distant dream. Add in analytics-based telemetry exposed on the disaggregated management controller and you have the foundation for a new logical architecture—a rack-level system.

This new logical architecture is available today in the form of an architecture that exposes a standard management framework via REST APIs to discover and retrieve the set of raw components—like drawers, blades, disks, and pooled resources like processors, memory, and NVMe disks—in a rack and collectively in a pod. These resources can be provisioned by a separate management network to compose a compute node or storage node.

In addition, a telemetry model is supported that exposes capacity, capability, and bottlenecks at each component level, thus allowing the right hosts to be composed for orchestrating a workload. Separation of the data and virtual management plane from the hardware provisioning and management plane and telemetry with analytics enables resources such as storage, memory, and compute to be added as needed, creating flexibility and scalability that can be fully utilized.

Of course, the success of this new logical architecture depends on the creation of open standards for the configuration and management of the rack components—such as compute, storage, network, and rack management controllers. These standards allow IT organizations to connect various hardware components together to form software-defined systems that more effectively utilize all the hardware in a disaggregated rack. They also allow for the architecture to evolve over time based on hardware innovation and changing customer use cases.

We pioneered the rack scale concept, working closely with key partners and standards bodies, such as DMTF Redfish. The players in these collective efforts recognized the importance of working with standards bodies to enable interoperable hardware, firmware, and software.

The resulting architecture is the result of an open effort that allows the industry to innovate and create diverse solutions to give customers many different choices. At the same time, the open approach establishes a platform for innovation at various levels in the data center market. It allows for the architecture to evolve over time based on hardware innovation and changing customer use cases.

Looking ahead, here is some of what we see on the horizon:

  • the ongoing disaggregation of compute, I/O, memory, and storage, which will give datacenter operators the ability to upgrade the different components independently of each other, everywhere in the datacenter
  • the evolution to disaggregated NVMe-as-storage solutions, pooled FPGA, and disaggregated networks delivered as solutions architected in rack scale
  • the development of an agile orchestration of hardware layer in open source solutions like OpenStack
  • the use of high-speed interconnections between components with less copper and more prevalent optical/wireless technologies, along with more security and telemetry at every level to drive more efficient use of resources

For more information on Intel® Rack Scale Architecture visit: www.intel.com/IntelRSA. And if you’re ready for a technical deep dive at this point, you can explore the details of Intel® Rack Scale Architecture firmware and software components on GitHub.


Partnerships Cloud Infrastructure

Mrittika Ganguli, Intel Cloud Platforms Group

Mrittika is a Principal Engineer and Platform Software Architect in the Cloud Platform Group, DCG – India, and has 19 years of experience. At Intel, Mrittika has worked on software systems design, development and architecture. Her work includes building the server system management software architecture for cloud and server hardware deployment for cloud in datacenters. She has been involved in OpenStack projects within Intel since 2010. She was a co-creator and architect for a OpenStack-based product for SLO management Intel(R) SAA. She has five patents granted and five filed and multiple published papers. Workload characterization, orchestration in clouds and disaggregated Rack architectures are her current areas of work.

Mrittika Ganguli, Intel Cloud Platforms Group

Discussions