PowerOne brings together the following networking components and technologies:
The Dell EMC PowerEdge MX platform is a next-generation modular platform that brings together compute, storage, and networking under a single management umbrella.
The PowerEdge MX7000 platform supports the following networking modules:
This section describes the internal architecture of PowerOne.
The MX scalable fabric architecture ties multiple chassis into a single network domain to behave like a single chassis from a networking perspective. It can scale up to 10 chassis, 80 compute sleds.
A scalable fabric consists of two main components — a pair of MX9116n Fabric Switching Engines (FSE) and MX7116n Fabric Expander Modules (FEM) used to connect remote chassis to the FSEs.
See Dell EMC PowerEdge MX7000 Technical Guide for more information about MX7000 networking.
The PowerEdge MX7000 infrastructure has numerous communication paths or fabrics interconnecting the modules. The management fabric connects from each management module (MM) slot with a point-to-point link to each front slot and each IOM slot. This fabric runs at 1 Gb/sec. A 1 Gb/sec link spans the two MM slots for communication between these modules.
The internal connection of the main I/O subsystem in a PowerEdge MX7000 enclosure is referred to as Fabric A. This fabric is connected using a direct orthogonal connection between sleds in front of the chassis and I/O fabric modules in the rear. The mating connection on the sleds is on the mezzanine cards.
The fabric enables two IOMs (A1 and A2) for redundancy. Each IOM connects to all eight compute nodes through the orthogonal connectors. The connection from the IOM to each compute node (the mezzanine card on the compute node) is an x8 bi-directional link. Depending on the interface, all the lanes within the link may not be used; however the x8 provides a path for future high-bandwidth fabric types.
The external port assignments of pass-through modules are fixed due to the direct routing connection internal to the system and the IOM itself. A standard size compute sled that is configured with dual port Fabric A mezzanine maps to a single port on each I/O module. Each compute sled number matches the IOM numbering on both the top and bottom I/O modules.
The following figure is an example of the MX740c mezzanine card slots.
The following figure shows all the compute sleds connecting to IOMs.
The IOMs A and B span the entire width of the chassis using direct orthogonal connections to the compute sleds. When installing these, the IOM will engage simultaneously with the Vertical Power Distribution Board (PDB) and the sleds.
The following figure shows PowerEdge MX7000 chassis with two scalable networking fabrics (Fabric A and Fabric B) and one redundant storage fabric (Fabric C). In the current PowerOne architecture, only Fabric A is used. Fabric B and Fabric C are not used.
The MX7000 I/O modules used are MX9116n Fabric Switching Engine (FSE) and MX7116n Fabric Extender Module (FEM).
The MX9116n Fabric Switching Engine (FSE) is a scalable, high-performance, low latency 25-Gbps Ethernet switch purpose-built for the PowerEdge MX platform for demanding compute and storage traffic environments.
Each PowerOne compute domain can use either two or four MX9116n IOMs. These MX9116 IOMs are installed in the first two chassis of a compute domain and provide the interconnection point between the LAN (leaf and spine), SAN, and up to ten MX7000 chassis.
The MX7116n architecture is extremely simple, consisting of a simple Ethernet repeater along with control circuitry for communication of low-level information (such as environmental information) to the MX7000 enclosure controller. Power and cooling is supplied by the MX7000 chassis.
PowerOne system fabric begins with a pair of Dell S5232F-ON (leaf) switches to support a single PowerOne Pod. The system fabric is designed for significant future scalability. In this design, an additional pair of S5232F-ON (leaf) switches will support each additional Pod. These S5232F-ON leaf switches will then connect to Dell Z9264F-ON (spine) switches.
For more information about these switches, see the following documents:
The Dell EMC PowerOne OOB management network consists of a pair of Dell S4148T-ON switches to support a single Dell EMC PowerOne Pod. The OOB network is designed for significant future scalability. In this design, the OOB management network will add a pair of Dell S4148T-ON switches to support each additional Pod. These S4148T-ON switches will then connect to Dell S5232F-ON (aggregation) switches.
For more information about these switches, see the following links:
The PowerOne SAN enables you to start with a small configuration, and then scale out to a larger configuration. The current per-Pod SAN configuration consists of two Connectrix MDS switches (9148T or 9396T).
The PowerOne SAN is designed for significant future scalability, to a core-edge design. In this design, the existing pair of SAN switches become edge switches, and a pair of Connectrix MDS switches (9396T, 9706 or 9710) are added to become the core. The edge will be expanded by adding a pair of edge switches for each additional PowerOne Pod. The core will be expanded by adding up to six additional core switches.
For more information about these switches, see the following link: