Dell PowerEdge MX7000 Chassis Review

Dell PowerEdge MX7000 Chassis Review

September 1, 2022 0 By Lorena Mejia

We finally got the Dell EMC PowerEdge MX7000, Dell’s replacement for the MX1000 enclosure (SHOP HERE). This is a 7U chassis with support for up to 8x single-width server sleds, four double-wide sleds or a combination of the two. If you will recall, the MX1000 blade server chassis is a 10U behemoth released in 2012 but the MX7000, even at 7U offers better performance and capabilities in a smaller space. It’s also designed to support up to 3x new CPU generations from both AMD and Intel, but so far only Intel Xeon Scalable processors are supported. It’s a cost-effective flexible architecture that’s easy to scale-out offering on-demand allocation of compute, storage and networking pools. 

Yes, this review of the Dell MX7000 enclosure is a little late in coming. Modular systems are not new to Dell as there are a few others still in the fold, although admittedly a little long in the tooth like the VRTX Tower and the 2U FX2 and FX2s platforms, plus the previously mentioned M1000e. The MX7000 at only 7U takes up a bit less space than the M1000e and doesn’t support as many blades. That said, technology has a way of making something better occupy a smaller area of space and the MX7000 does just that.

Here are some direct comparisons between the two enclosures the old M1000e and the MX7000. As you can see, some similarities like the same number of I/O modules, same number of power units and power draw but that’s about it. One of the main differences on the MX7000 enclosure compared to the M1000e is no midplane. That was also one of the limiting factors on the M1000e. Instead, the server nodes on the MX7000 connect directly to the I/O, and storage modules for what Dell calls a “Kinetic infrastructure” for the modern software-defined data center. In other words, quick and agile scaling of resources. This means no replacing the midplane board when new technologies become available. 

Taking a look at the front of the system, there are 6x hot-swappable platinum rated 3000W PSUs across the bottom. Above those, a central bank of 4x 60mm fans vertically aligned that separate the server nodes into 2 groups with bilateral symmetry. You can load 8x single-width chassis, 4x double width chassis, or a combination of the two. There is even talk of half-height modules in the future so perhaps up to 16 of those. There are currently only 4 modules supported on this system. 

Two single-width compute modules, the 6-bay, dual-socket MX740c with support for Intel Xeon Scalable processors from gen 1 and 2. The MX750c, also 6-bays but with support for 3rd generation Intel Xeon Scalable processors. Next a double-width, quad-socket server sled, the 8-bay MX840c, with support for 1st and 2nd generation Intel Xeon scalable processors. Both the MX740c and MX750c can provide almost 8TB of memory at capacity, while the MX840c can deliver a little over 15TB of memory. All can be outfitted with Persistent memory modules, but only the MX840c and MX740c can host non-volatile DIMMs, and only with the 1st generation Intel Xeon Scalable CPUs. NVDIMMs will reduce the number of available drive bays because of the additional battery required to protect the DIMMs during a power loss. Lastly, there’s a single-width MX5016s storage module supporting up to 16x hot-swap 2.5-inch SAS drives. You can install 1x to 7x MX5016s storage modules for up to 112 drives, but you will need at least one of the compute modules. 

Control panels are mounted in both the left and right server ears of the Dell MX7000 blade chassis. The left server ear has 3x configurations. LCD panel with QuickSync, and an LCD panel with no Quick Sync. That last would be the one we have on this system. On a side note, quick sync does have to be ordered at time of purchase and we can do that at, we are a partner with Dell after all…

Both of those control panels have LEDs, with a system ID button LED indicator, an options button for network settings and system information, next, the optional quick sync indicator, then a system alerts indicator followed by an LCD activation button, and the Quick Sync wireless status indicator. There’s also one with just the LED indicators showing; system health, system temperature, I/O module health, fan health, stack or group and LED status bar, and a system ID button. Quick Sync is a nice feature for at chassis management of the system using an iOS or Android smart phone or tablet. 

The right server ear control panel is a little simpler and offers a power ON button, 2x USB 2.0 ports and a mini-display port, plus a management port squeezed in between.

We’ve already reviewed two of the server nodes for the MX7000, the Dell EMC PowerEdge MX740c and the Dell EMC PowerEdge MX840c. The MX750c was not released when we did those reviews, so let’s take a quick dive into the features on that system. Hopefully we will get one of those in soon but here is the quick overview. 

Dell added the PowerEdge MX750c to the mix about a week after Intel’s announcement of the 3Rd generation Xeon Scalable processors in March of 2021. 3rd gen Xeon Scalable processors offer 8x to 40x physical cores and 16x to 80x virtual threads. A significant boost from the 28 cores available on the 1st and second-generation Xeons Scalable processors. They also support faster and more memory with 8x memory channels instead of 6, plus PCIe 4.0 for improved performance with I/O and NVMe storage devices. There is a configurable TDP limit of 270W on these CPUs. Keep in mind you will only get that top core count using the Platinum processors. The larger socket design is also incompatible with the 1st and 2nd generation CPUs. the MX750c does support up to 6x hot-plug SAS, SATA or NVMe storage devices at up to 7.68TB each. NVMe drives can be plugged directly into the system board with 3x PCIe Gen4 x8 connectors and 2x drives each. A dedicated PCIe 4.0 x16 socket for a PowerEdge RAID controller is available for support of SAS or just more control over a SATA implementation.  

You can install up to 7.68TB in 32x DIMM slots of memory using 16x 200 series Persistent memory modules, PMem, paired with 16x RDIMMs at full capacity. Or up to 2TB with just RDIMMs or 4TB at capacity with LRDIMMs. Memory speed is dependent on the CPU tier with Gold and Platinum providing 3200MT/s while Silver will only get you up to 2666MT/s. Aside from large performance gains as a result of the 3rd generation Intel Xeon Scalable processors, this platform has remarkably similar layout to the MX740c. It also supports a dedicated slot for a boot optimized subsystem, BOSS, with dual M.2 drives or a dual SD card module (DSDCM) but not both. Physically impossible since they would have to occupy the same space. 

RAID options for the compute modules include the S140 or the S150 a software RAID option that’s built-in. A few of those controllers are designed for the MX750c PCIe 4.0 interface. There are quite a few options for support of both local and external drives. A mini mezzanine connector plugs into the storage Fabric C module for a seamless SAS fabric connection without any cables. Let’s look at the back of the MX7000 chassis and see how it all connects. 

Everything on the back of the system is horizontally aligned, while everything on the front of the system is vertically aligned. The allows the compute or storage module to straddle multiple I/O and management modules in an orthogonal, or right-angle fashion. The MX740c, MX750c and MX840c can host 3x fabrics, through the I/O modules in back with 2 pairs of modules for redundancy. Assuming you go with full redundancy you can have two fabric A modules, 2x fabric B modules and two Fabric C modules.

The management modules are also redundant and you could go with two of those as well. There are several options for the A and B fabric modules offering different ports and connection speeds. Fabrics A and B are the main I/O system in the MX7000 chassis. Each of those modules can connect to all 8x compute sleds by connecting through the mezzanine cards.  

The Dell EMC PowerEdge MX5016s storage sled, which connects through Fabric C can be outfitted with up to 16 SAS storage devices connected at 12Gb per second through an MX5000s fabric C I/O module which mounts in back on the MX7000. The MX5000s is a redundant hot-swappable switch solution that plugs in at the back of the chassis and can be connected to several compute sleds and MX5016 storage sleds both in this chassis and remotely in other chassis. 

I/O mezzanine cards at the rear of the compute sleds have dual ports for redundancy connecting to I/O Fabrics A and B. Of course, on the MX750c you have a PCIe 4.0 bus for twice the speed of PCIe 3.0. A third PCIe slot hosts a mini-mezzanine slot with 2 x4 PCIe paths for connecting to each of the two storage Fabric C I/O modules. The I/O Fabric modules can be configured for different network connection speeds using different modules. Same for the pair of storage specific bays for Fabric C. Fabric C is currently for use in SAS or Fibre Channel storage connectivity. 

The MX5016s storage sled uses Fabric C to communicate directly with the I/O modules. Drives in the storage sled can use drive-assigned mode, which connects only to a single compute sled and can be assigned any disk quantity or order. Alternatively, Enclosure-Assigned mode maps an entire storage sled to a compute sled. The lack of a backplane means to support future technologies, all that will be required is to replace the I/O module for fabric A, B or C, and replace the I/O mezzanine cards in the sleds, if necessary. The server sleds on the other hand do still have storage backplanes for the upfront drives and there are also a few of those depending on your storage requirements for the specific compute sled. 

The MX7000 offers embedded, agile management of all components. Two redundant MX9002m management modules support the embedded management solution, the Dell OpenManage Enterprise Modular Edition or OME. The compute sled level each is equipped with the integrated Dell Remote Access Controller or iDRAC 9.0 modules that connect to the MX9002m management modules for comprehensive coverage of individual, multiple servers, and multiple chassis. QuickSync II which uses the OpenManage Mobile app also provides options for at chassis management of the system. On another note, a Dual SD card module can also be used for a hypervisor and features two SD cards on one side and a Flash drive on the other. That flash drive can be used by iDRAC to store system firmware updates and patches.The MX7000 blade enclosure is a worthy addition to data centers, cloud, and Enterprise businesses offering a highly adaptable and composable architecture that can quickly scale and be reconfigured to meet current demand, and demand into the future. With support for the next three Generations of processors from Intel, and one would hope that will include AMD EPYC processors soon, this platform will easily provide that return on investment! Upfront costs may be high, but if you have a large business, it’s worth the expenditure.

IT Creations carries this enclosure, and more, and we have the blades to outfit it too, so check out ITC!