Home > Storage > PowerMax and VMAX > Storage Admin > Design Guide—VMware IaaS with Big Cloud Fabric on PowerEdge MX Servers and PowerMax Storage > Ready Stack architecture with Big Cloud Fabric
Dell EMC PowerEdge MX7000 is a modular chassis infrastructure that accommodates various compute and storage sled combinations. High-speed fabrics connect compute and storage, which share power, cooling, and OpenManage Enterprise Modular Edition systems management. The chassis features two-socket and four-socket servers with Intel Xeon Processor Scalable processors and up to 24 or 48 DIMMs, 12 Gb/s SAS storage sleds, and a choice of networking-switch I/O modules.
The following figure shows the front view of the MX7000 infrastructure. From left to right are four 2-socket and one 4-socket blades, and two storage sleds.
Figure 2. PowerEdge MX7000 front view
The following figure shows the rear view of the infrastructure.
Figure 3. PowerEdge MX7000 rear view
PowerEdge MX series infrastructures have the same foundational elements and benefits of the rest of the PowerEdge portfolio, providing:
This architecture provides for optimal utilization and flexibility, with maximum productivity and availability.
The MX7000 modular chassis and its compute, storage, and networking components create a portfolio that can support many types of workloads. It is well suited for workloads that require flexibility in assigning ratios of compute, storage, and networking, and the ability to easily change those ratios as workload demands change.
The MX7000 is designed to optimize application performance and ensure a stable environment, with intuitive tools that enable simplification and automation throughout the entire server life cycle. The MX7000 provides the flexibility that is required to build a scalable infrastructure that is tightly integrated with VMware vSphere.
The following table summarizes the MX7000 features.
Table 1. MX7000 modular chassis features
Feature |
Description |
Compute compatibility |
MX740c/MX840c |
Form factor |
7U |
Number of blades |
Up to 8 x 2-socket and 4 x 4-socket |
Local blade storage (optional) |
|
Local chassis storage (optional)
|
Up to 16 x 2.5-in. DAS drives occupying one slot |
Fabric types supported |
Up to 25 GbE, 32 Gb/s FC, 12 Gb/s SAS |
Power supplies |
Up to 6 x 3,000 W hot-plug PSUs (PSU and grid redundancy support) |
Management modules
|
Up to 2 redundant management modules |
KVM options |
Integrated with management processor |
Embedded in every PowerEdge server is iDRAC9 with Lifecycle Controller. iDRAC9 provides secure and remote server access for a multitude of common management functions. iDRAC with Lifecycle Controller operates regardless of the operating system state or the presence of a hypervisor. It offers a complete set of server management features including configuration, operating system deployment, firmware updates, health monitoring, and maintenance.
iDRAC9 provides various remote connectivity interfaces and protocols, enabling administrators to securely configure, deploy, manage, monitor, and update the server. IPMI, Redfish, SMASH-CLP, and WS-Man are a few of the common standard management interfaces that iDRAC9 supports.
The iDRAC9 HTML5 web UI provides for secure connectivity through HTTPS. With the Group Manager feature in iDRAC9 Enterprise, administrators can create a one-to-many console experience, which enables them to view and manage a set of servers rather than visually inspecting servers for faults and manually managing the servers. Combined with OpenManage Enterprise Modular Edition, iDRAC9 enables multichassis management in a one-to-many interface.
Network design is a fundamental aspect of the Ready Stack architecture and consists of these functional groups:
The primary building block of the OOB network is the Dell EMC PowerSwitch S4148T-ON switch, which provides 48 ports of 1/10 GbE and multiple uplink port options (10/25/40/50/100 GbE).
The key building block of the LAN is the Dell EMC S5248-ON switch. The switch provides up to 128 ports of 10/25 GbE, 32 ports of 40 GbE, 64 ports of 50 GbE, and 32 ports of 100 GbE. Some port configurations require breakout cables. Using two edge S5248F-ON switches provides redundancy. Integrated NICs directly connect the management hosts, which are rackmount servers, to the edge switches.
The platform and compute hosts, which are blade servers that are installed in the MX7000, are a different case. Inside the MX7000, a unified connection carries both LAN and SAN traffic, which must be separated before leaving the MX7000 chassis. The separation occurs in the Dell EMC MX9116n Fabric Switching Engine (FSE), which is installed in the back of the MX7000 chassis. The FSE connects to both the LAN (at the edge switch) and the SAN. Two MX9116n FSEs, one per MX7000 chassis, provide redundancy in a two-chassis configuration. Also, the Ready Stack includes PowerEdge MX7116n Fabric Expansion Modules (FEMs)—one per MX7000 chassis in a two-chassis configuration—to provide blade-server connection redundancy.
The key building block of the SAN is the Dell EMC MDS Series FC switch. The SAN can be configured as a core-edge or edge-core-edge design, depending on requirements. For the models used in our Ready Stack configurations, see Design configurations and specifications.
BCF from Big Switch Networks brings hyperscale data center design principles to enterprise data centers. BCF is powered by spine and leaf fabric architecture that provides physical network automation, visibility, and troubleshooting for VMware environments.
BCF provides application agility due to automation, operational simplification due to SDN, and significant cost reduction due to hardware/software disaggregation that is enabled by Dell EMC Open Networking switches.
Combined with the high-performing Open Networking switches, BCF provides network automation for multiple VMware products. BCF delivers new visibility and troubleshooting capabilities for both VMware and network administrators, with the aid of the BCF plug-in for vSphere Web Client and VMware vRealize Log Insight.
BCF lets you manage the network infrastructure as one big virtual switch compared with traditional box-by-box management, as shown in the following figure. Having one logical switch simplifies and accelerates the administration and management of the networking infrastructure.
Figure 4. One logical switch
The Dell EMC PowerMax family of storage systems provides simplicity, modern design, and flexible deployments. PowerMax systems implement an integrated architecture for block, file, and VMware Virtual Volumes, with concurrent support for native NAS, iSCSI, and FC protocols. The PowerMax systems support file and block environments, point-in-time snapshots, thin clones, synchronous and asynchronous replication, integrated encryption, tiering to the cloud, and deep ecosystem integration with VMware, Microsoft, and OpenStack.
The PowerMax family consists of two models, as shown in the following figure. The PowerMax 2000 provides efficiency and maximum flexibility in a 20-U form factor. The flagship PowerMax 8000 is designed for massive scale, performance, and I/O density, all within a two-floor-tile footprint.
Figure 5. PowerMax platforms
Both PowerMax models have at their foundation the Dynamic Virtual Matrix architecture and the PowerMaxOS 5978 operating system for the NVMe platform. PowerMaxOS can run natively on both PowerMax systems and, as an upgrade, on legacy VMAX All Flash systems. As with the previous generation of VMAX All Flash systems, PowerMax systems are all-flash arrays, which are targeted to meet the storage capacity and performance requirements of the all-flash enterprise data center.
PowerMax systems take advantage of the high-capacity NVMe flash drives that are used in the densest configuration possible. PowerMax platforms offer enterprise customers trusted data services, along with the simplicity, capacity, and performance that their highly virtualized environments demand. Yet these platforms still meet the economic needs of more traditional storage workloads. Also, with PowerMax systems, you can deploy applications that demand lower storage latency and higher I/O densities, such as real-time analytics, machine learning, and Big Data.
PowerMax systems offer these primary benefits:
Dell EMC Unisphere storage management is an HTML5 web-based application that delivers the simplification, flexibility, and automation that are required to accelerate the transformation to the modern data center. For customers who frequently build up and tear down storage configurations, Unisphere for PowerMax makes reconfiguring the array even easier by reducing the number of steps that are required to delete and repurpose volumes.
PowerMax systems deliver deep integration with all the VMware vSphere core storage-enablement primitives. vSphere Storage APIs – Array Integration (VAAI), the core VMware API for storage offload, significantly increases the performance of the vSphere cluster. PowerMax systems support all VAAI primitives—XCOPY, UNMAP, WRITE SAME, and ATS Lock Extents. The systems also support VMware Virtual Volumes and vSphere APIs for Storage Awareness (VASA) 2.0 for next-generation VMware environments.
PowerMax systems provide additional integration points, beyond the core VMware integration points, to ensure a complete integrated user experience that is familiar to the VMware administrator. Dell EMC VSI plugs into vSphere Web Client, enabling administrators to directly provision PowerMax storage, visualize the infrastructure, manage local replication, schedule UNMAP operations, and more. The Dell EMC Storage Analytics (ESA) plug-in provides integration with VMware vRealize Operations Manager to provide a complete end-to-end, in-context view of systems and storage health. Further, the Dell EMC VMAX and PowerMax Content Pack for vRealize Log Insight provides dashboards, alerts, and queries to sort and intelligently present PowerMax log information directly from vCenter. All PowerMax integrations for VMware environments are available at no additional cost.
This Ready Stack design uses a three-server external management cluster and, internal to the MX7000 chassis, a platform management cluster. This design follows the VMware Validated Design (VVD) Standard SDDC and provides maximum security, resource isolation, and scalability.
In this Ready Stack design, the management cluster consists of three R640 rack servers. Having a separate management cluster of hosts, rather than a single host, ensures that management components have dedicated resources that are isolated from the production compute environment. Having a separate management cluster also provides for independent scaling of compute resources between management and production workloads. This separation makes it easier to determine the management components and their related infrastructure, reduces resource contention, and provides physical security and process boundaries.
The management cluster is configured to support a single host failure, commonly referred to as N+1. Adding management components beyond those that are specified might require additional hosts. The BCF Controller is a VM that is hosted within the management cluster. The version of vCenter running in this cluster also includes the BCF plug-in for vSphere Web Client.
The production (compute) cluster resides on blades in the PowerEdge MX7000 chassis and contains the IaaS components that are described in Infrastructure as a Service. The platform management hosts provide the automation and content catalog resources that are needed to support an IaaS production environment. This cluster is separate from the management cluster to ensure the performance, scalability, and security of the management and production clusters.
Separating the management and production clusters from one another also achieves a greater level of resiliency. When designing larger environments that use multiple MX7000 chassis, disaggregate the production cluster and spread it across multiple chassis and racks, as shown in the following figure:
Figure 6. Platform management on multiple MX7000 chassis
This disaggregation improves the fault tolerance of the cluster by removing potential points of failure. You can apply this method to other clusters as well to improve their resiliency.
Using the BCF Controller to perform configuration, automation, and most troubleshooting tasks dramatically decreases the number of management consoles that are required to provision new physical capacity or new logical applications. For example, in a 16-rack pod with dual leaf switches and two spine switches, a traditional network design would have 34 management consoles. The BCF design has only one—the controller console—to perform the same functions. The result is massive time savings, reduced error rates, and simpler automation designs.
As shown in the following figure, BCF Controller provides a single view for assessing the overall network status:
Figure 7. BCF UI dashboard page
In traditional network administration, the operational workflows are mostly manual (sometimes scripted, but the scripts must be maintained), resulting in loss of agility. The BCF architecture has integrated network automation. Rapid network upgrades are up to 20 times faster than legacy networks, resulting in lower network downtime, higher agility, and significant savings in operating expenses.
Integrating BCF with the VMware SDDC provides:
BCF integration with VMware vSphere, NSX, vSAN, Integrated OpenStack, and vRealize Log Insight provides network automation and visibility. Network administrators gain visibility into the virtualization environment through the BCF Controller and can use the BCF plug-in for vSphere Web Client to provide fabric visibility to the VMware administrator.
This section briefly describes the IaaS components of this Ready Stack.
VMware IaaS provides virtualized computing resources by separating applications and their resource demands from the underlying infrastructure resources on which they rely. VVD for SDDC extends the typical IaaS solution to include operations management, business continuity, and security.
The combination of compute, storage, and networking hardware in this Ready Stack provides an ideal platform on which to configure IaaS. For more information, see Validated Designs on the VMware website.
Dell EMC ESA for vRealize Operations enables customers to optimize performance and diagnose issues across their physical storage and VMs. Dell EMC ESA for vRealize Operations:
Through log analytics, aggregation, and search, vRealize Log Insight delivers automated log management that provides operational intelligence and visibility for enabling service levels and increasing operational efficiency.
The Dell EMC VMAX and PowerMax Content Pack, when integrated with VMware vRealize Log Insight, provides dashboards and user-defined fields for VMAX and PowerMax storage. These tools enable administrators to conduct problem analysis and analytics on their arrays.