Dell Technologies Kinetic Infrastructure: On the Path to Fully Composable Infrastructure

Co-Authors:

Jon Hass, Senior Distinguished Engineer, Server Infrastructure Solutions, Office of CTO

Sudhir Shetty, Distinguished Engineer, Server Infrastructure Solutions, Systems Management

Where we are on the path

Over the last several years, we have written about composable infrastructure and our vision for delivering full composability. A few samples of these writings can be found here and here. We thought now would be a good time to give an update on some concrete steps we have taken on the road for delivering our vision.

Kinetic Infrastructure is the term we use to embody both our composable architecture vision and the real embodiments of technology we deliver on the way. Gartner recently released a report, “Understanding the Hype, Hope and Reality of Composable Infrastructure,” commenting on the industry state of composable infrastructure. In their report, Gartner lists many of the customer benefits of composable infrastructure. Their list, which we believe, matches closely to kinetic infrastructure benefits of which we have previously blogged, includes increased efficiency in resource utilization, increased flexibility in server configurations, and enabling hardware as a service capability.

Gartner’s report also lists several items that need to be addressed to realize the full potential of composability. We believe two of these items are the need for additional management software for automation and the need for standard APIs.

To address the need for standard APIs for composability, kinetic architecture is utilizing the mechanisms of DMTF Redfish and the principles described in the Redfish Composability White Paper. Dell Technologies have been leaders in the development and adoption of the Redfish standard and our last two generations of servers have enabled the Redfish APIs for server management. We believe that Redfish’s RESTful API is the best mechanism for allowing the interaction between higher level workload orchestration software layers, like VMware Cloud Foundation™, and composable hardware infrastructure. Through tight integration with orchestration layers, we can address the need for automation of the composable infrastructure management.

The path to delivering the kinetic architecture journey: MX7000 and VMware Cloud Foundation

In late 2019, Dell Technologies and VMware released composability integration between VMware Cloud Foundation and Dell EMC PowerEdge MX. VMWare Cloud Foundation (VCF) is an integrated software stack that delivers a faster path to a true integrated hybrid cloud. It automates the deployment and lifecycle management of a software-defined data center on standardized hyperconverged architecture and offers a consistent operational model based on well-known vSphere and vSAN tools and processes.

PowerEdge MX platform provides a kinetic infrastructure of compute, storage and networking resources that can grow with customer workload demands. A single PowerEdge MX7000 chassis has 8-slots that can be used for compute sleds or storage sleds. A storage sled has 16 SAS drives each of which can be assigned to a compute sled. In a multi-chassis group, there can be up-to 20 chassis that are managed from the lead chassis.

The Management Module embedded in the chassis hosts the Redfish Composition Service that permits a client to discover the available resources within the platform, understand the assignment constraints and perform the assignment of resources to compose a server. This functionality is available as soon as the hardware is brought online without requiring any additional software layer to be installed, which simplifies and minimizes the overall deployment and setup time for customers who are deploying VMware Cloud Foundation.

Figure 1 VMware Cloud Foundation/MX integration

The MX7000 Redfish Composition Service has 3 key constructs.

  • Resource Blocks -the building blocks of hardware resources that can be composed by an external client to create a full-fledged computer system.
  • Resource Zones – are used to represent the constraints to the client on what resource blocks can be composed together. So, resources within the same resource zone can be composed together.
  • Composition Requests –The client can evaluate the workload or user request for a computer system and evaluate the zoning and block information to create a composition request. The client can a) add/remove resource blocks to an existing computer system or b) create a composition request of the desired computer system.

The PowerEdge MX firmware supports resource block types of Computer System and Storage and add/remove resource block requests.

VMware Cloud Foundation supports the notion of composable infrastructure by interfacing with a Redfish end-point that supports the Redfish composition service. It provides a flexible interface that permits the user to define the type of computer system desired. The user can filter for the type of computer system by specifying the range of resource values (e.g. CPU cores, CPU speed) and storage required to fulfill a workload need. The VMware Cloud Foundation composability layer can create a composed server by allocating the appropriate storage drives to a compute sled.  Once composed, a server can be commissioned to be used in a workload domain.

The VMware Cloud Foundation interface also allows the decomposition of a composed system (which removes all assigned resource blocks such as storage) or add/remove storage capacity.

The path ahead

Gartner says “Today, composable infrastructure is held back by a lack of technology to disaggregate DRAM from processors, industry-standard configurable fabrics and cross-vendor APIs.[i]” We believe the best way to address the cross-vendor API issue is through Redfish. The work Dell Technologies and VMware have done shows how we can manage composability through workload orchestration software utilizing standard APIs. Navigating the path forward and addressing Gartner’s other concerns which we believe requires expanding the mechanisms we developed with the MX7000 and VMware Cloud Foundation beyond SAS storage composition. Industry efforts such as NVMe-over-Fabrics (NVMeOF) will allow the disaggregation of NVMe storage devices. Other industry efforts, such as Gen-Z, will allow the disaggregation of memory devices, like storage class memory (SCM), and accelerators. Dell Technologies’ kinetic infrastructure with its utilization of industry standard Redfish APIs can adopt new technologies as they mature, expanding the types of resources that can be composed.

If you are attending Dell Technologies World 2020 be sure to attend the breakout track, “Bringing Cloud Automation and Agility to Your Data Center with Dell Technologies Cloud,” where we will discuss how a software-defined data center approach with kinetic architecture and VMware Cloud Foundation can provide customization and flexibility for bridging your data center operations on the ground into SDDC and cloud. And stop by and visit us at the Solutions Expo, we will be talking and showing kinetic infrastructure in the Dell Technologies Infrastructure Solutions Group booth.

Visit our website for for additional information about PowerEdge servers and be sure to follow us and join the conversation on Twitter @DellEMCServers.

[i] Ibid

About the Author: Bill Dawkins

Bill Dawkins is a Fellow/VP in the Dell EMC Server Solutions Group, Office of the CTO. Bill has 24 years of technology industry experience and has been with Dell since 1999. During his time at Dell, Bill has worked on both server and storage technologies and has covered areas such as software defined storage, hyper-converged systems, object storage, storage area networks and storage caching. His current areas of investigation include Memory Centric Server Architecture and Server Disaggregation. Bill holds a Ph.D. and an M.S. in Computer Engineering from Rice University and a B.S. in Electrical Engineering from Oklahoma State University. He has 29 issued patents and 5 patents pending.