Dell EMC will launch their new modular PowerEdge MX platform September 12, promising to remake data centre architecture with technology that will support both traditional and next-gen applications within a composable computing framework that pays much more attention to memory than earlier iterations of that technology.

In what Dell EMC is calling the most substantial PowerEdge announcement they will make this year they have unveiled PowerEdge MX, which has a modular architecture designed to handle both traditional workloads and next-generation workloads like in-memory computing. It features a modular architecture designed to support both today’s mainstream data centre applications, as well as the cutting edge ones and the applications of tomorrow. The announcement comes in the run-up to next week’s VMworld event, where the PowerEdge MX will be demoed for the first time.

News of the PowerEdge MX will not take the industry by storm.  Jeff Clarke, Dell’s Vice Chairman, Products & Operations, previewed it at Dell World this spring. Now, however, with the product less than a month away from general availability, more details are available.

“The most exciting part of this announcement is the modular architecture that helps build a bridge between traditional workloads and transformational workloads, which include in-memory, deep learning, and machine learning,” said Ravi Pendekanti, SVP Product Management and Marketing, Server and Infrastructure Systems at Dell EMC. “The challenge for organizations is both running workloads of today AND the workloads of the future, and that’s where this comes in. It runs them both in a single infrastructure. A typical use case would be running a traditional SQL database for an OLTP application while running SAP HANA for in-memory – and running both of those workloads at the same time.”

Pendekanti said that PowerEdge MX also fits into the concept of composable infrastructure, where all the elements of data centre infrastructure can be broken down, disaggregated and pooled. Dell EMC sees the concept of kinetic infrastructure as central to achieving this goal.

“Composable infrastructure lets you pick and choose what you need and build the right infrastructure to run your app,” Pendekanti said. “To date, however, it has been very compute-centric, and lots of other resources are unused and wasted. The goal behind kinetic infrastructure is very simple – to unleash energy and use these resources to optimal capacity.”

The third element of the announcement is Open Manage Enterprise – Modular Edition, which Pendekanti said will be able to better manage this blend of old and new workloads to get better utilization of things like memory which have been poorly served to date.

“The issue is how do you get resource utilization higher than 50 per cent, where it has been,” he stressed. “The goal is not to do a forklift upgrade, replacing the old infrastructure with an accompanying migration.  Kinetic infrastructure enables the seamless ability to run multiple workloads, both traditional and new. It builds products to be memory-centric – very different from the old compute-centric model – to get rid of bottlenecks to optimal performance. The PowerEdge MX talks to this exactly, with a very adaptable infrastructure that is nimble and flexible.”

“The PowerEdge MX is the first instantiation of a modular solution that will deliver on our vision for a kinetic infrastructure, help unbound IT organizations and give them the flexibility to drive innovation,” said Brian Payne VP Product Management and marketing, Dell EMC PowerEdge. The design principles he indicated that the company is emphasizing to the market is the ability to deliver a flexible architecture to handle both the  uncertainty of the workloads of the future with the certainty of existing ones, agile management to make the IT organization more effective to serve the needs of the business, and a responsive modular design to bring compute, storage and networking together and keep up.

“At launch on September 12, in addition to the Dell EMC PowerEdge MX7000 7U 8-bay chassis, we will have two compute sleds, the two socket MX740c and he four-socket MX840c,” Payne said. “We have integrated industry-leading local DAS storage in each sled, with up to six NVMe, SAS or SATA drives supported in the two-socket sled.” The MX840c can support eight drives.

“In addition to the leading amount of storage per compute sled, it can scale further with the Dell EMC PowerEdge MX5016s storage sled, which slides into the MX infrastructure,” Payne added. The storage sled holds up to 16 hot-pluggable SAS storage hard disk drives, with a maximum of seven MX5016s sleds in the MX chassis for up to 112 drives of direct-attached storage, to accommodate specific use cases.

New MX Ethernet and Fibre Channel switching modules are part of the system as well, making the PowerEdge MX the industry’s first modular infrastructure to deliver end-to-end 25GbE and 32Gbps fibre channel host connectivity.

“This is a critical element of the MX offer,” Payne said. “Our vision of a disaggregated server changes the way all components in the data centre  connect together, with rapid change coming from 25 GbE to 100 GbE, and ultimately to 400 GbE and whatever comes beyond that. Others have midplane designs which limit the ability to change to adapt to this. This has no midplane at all. It means they can adapt as connectivity changes over time without the need to rip and replace. We have also thought about the cooling, which will be extensible into future technology.”

Dell EMC also announced Dell EMC OpenManage Enterprise – Modular Edition, which adapts the functionality of OpenManage Enterprise management within this modular environment.

“It provides a unified solution for all form factors, with a unified simple interface, for simplicity in managing the data centre environment,” said Drew Schulke, VP Product Management, Dell EMC Networking. “We also made a couple of deliberate design decisions. We support a high amount of bandwidth between nodes, with 25GbE from day one, and we have a highly scalable fabric under a single switching infrastructure.”

“This is ideal for service providers, who are interested in streamlining operations to better meet end user needs,” Payne said. “This automated management of their infrastructure greatly simplifies things and can manage up to 10 chassis with a single fabric.”

Channel service providers and Cloud Service Providers will be a key part of the go-to-market, as well as VARs, Pendekanti said.

“The enablement of partners is happening now, on both the VAR and service provider/CSP side,” he indicated.