Dell EMC, AMD
Dell EMC launched its Advanced Micro Deviceslineup in China today, a few months after announcing its reengagement with AMD . This is an important milestone for both Dell EMC and Advanced Micro Devices. For Dell EMC, it is a furthering of the company’s strategy of delivering choice in the marketplace. For AMD, this launch is another validation of its datacenter strategy and its progress. Dell EMC is targeting these servers at edge computing, software-defined storage, HPC, and virtualization. I see these as three general-purpose servers that can be scaled out or scaled up.
What Dell EMC launched
The AMD lineup consists of the following servers:
Dell EMC PowerEdge line up powered by AMD EPYC
- Dell EMC PowerEdge 6415 – a 1U single socket server designed for dense computing. This server will support up to 32 cores, 2TB of DDR4 RAM, and 10PCIe NVMe drives. When I think of the 6415, I think of scale-out virtualization and software-defined storage (SDS). The 1U form factor should enable highly dense, highly performant virtualization environments.
- Dell EMC PowerEdge 7415 – a 2U single socket server designed for software-defined storage. Like the PowerEdge 6145, this server will support up to 32 cores and 2TB of DDR4 RAM. However, the 7415 will support up to 24NVMe drives. Given the high number of NVMe drives, this server should support scale-out SDS environments well. It’s interesting to note that the PowerEdge 7415 is the first AMD-based server that has been VMware vSAN certified.
- Dell EMC PowerEdge 7425 – a 2U, two-socket server that should excel in certain HPC workloads. With support for up to 4TB of DDR4 RAM across 16 memory channels and 128 PCIe lanes, the 7425 should also be a virtualization beast.
Dell EMC traditionally has a three-digit numbering scheme to name its PowerEdge servers. The higher the first digit, the more powerful the server. Not so with the EPYC powered servers. One of AMD’s claims is dual-socket performance in a single socket server (more on this in a little bit). The “1” or “2” indicates how many sockets are in the server.
Enterprise-grade single socket server?
Dell EMC bought into the AMD “single socket” value prop and it's a good one-two of the three server configurations. For those wondering what that single socket value prop might be, consider the following:
- Even with virtualization, the average two-socket server (with both sockets populated by a CPU) peak at about 30% (CPU) utilization.
- About 1/3 of two-socket servers ship out of OEMs with only one socket populated (with a CPU).
In other words, IT organizations are far from compute efficiency and are wasting money. One could speculate that two-socket servers are purchased for reasons other than compute (i.e. memory, PCIe slots, storage, etc.). Otherwise, CPU utilization would be much higher. So, a “tax” is essentially being paid for access to those resources associated with a second CPU. Another potential explanation is that IT organizations opt for the second socket in the interest of “future proofing”. As an ex-IT Director, I can’t recall ever opening a server and installing a second CPU. Ever.
Regardless of such speculations, purchasing a two-socket server with only one socket populated is a waste of money. So, why not offer a single socket enterprise-class server and allow organizations to recognize real cost savings? This seems to be AMD’s thinking around its single socket value prop, and it appears Dell EMC has taken a market leadership position with the new PowerEdge 6415 and 7415. AMD's single-socket is brilliant.
Analyst note: I believe the reason for such low utilization in virtualized environments has more to do with an imbalance of resources. That is, too many cores accessing not enough memory across too few channels. A balanced microarchitecture like EPYC will lead to greater utilization, regardless of socket count.
Are things any different this time around?
When Dell (pre-EMC) first launched an AMD Opteron server platform back in 2006(ish), it kind of felt like a “me-too” launch. There didn’t seem to be a lot of thought around design, positioning, or marketing of the AMD portfolio. If a customer asked, great. But Dell didn’t go too far in trying to distinguish between AMD and Intel other than the “5” or “0” at the end of the product model number.
This time feels different to me. When it comes to AMD and Intel
, there’s always going to be some element of “me too”. This is unavoidable. This time around, though, Dell EMC seems to have looked at where EPYC shines and built platforms in support of workloads that can best take advantage of the EPYC architecture. Take the PowerEdge 7425 as an example. Because the EPYC architecture excels in floating point performance, Dell EMC designed a server platform around EPYC that would fully exploit these capabilities for workloads like computational fluid dynamics (CFD)—lots of PCIe slots for GPUs and accelerators and lots of high performing storage. Principal analyst Patrick Moorhead and I will be looking to see how far Dell EMC leans into AMD's server chip this go-around.
Dell EMC has taken a smart approach with its AMD-based PowerEdge portfolio by being targeted in positioning and broad in availability. This allows Dell EMC to draw distinctions with competitors (and even its own Intel-based portfolio) while letting the market ultimately decide where these servers will find traction.
AMD EPYC has found success in cloud and hyperscale environments, and I’m sure Dell EMC will ride this momentum. Where Dell EMC can grab a market leadership position is in channels. The success of the AMD portfolio in Dell channels will directly correlate to the investments made by both Dell EMC and AMD. Campaigns around education and awareness, tied to incentive-based motivation for sales reps at CDW, SHI, and Connection will ultimately bear fruit. The question is whether Dell EMC or AMD have the funding and patience.
What about the EPYC security?
If there was one area where I think Dell EMC could have been stronger, it’s the integration of security between AMD’s EPYC and PowerEdge’s silicon root of trust. Both AMD and Dell EMC have built-in technologies that tighten server security. Dell EMC enables a silicon root of trust to better protect against rootkit and firmware attacks. AMD EPYC will eventually extend security from server boot to spinning up virtual machines, to cloud migrations, to server shut down. Dell EMC should consider how to seamlessly integrate these technologies.
Dell EMC keys to success
While I’m not a fan of the “if you build it, they will come” product marketing strategy, I believe there is an element of this in play. There is a pent-up demand for EPYC and Dell EMC is delivering a number of platforms that can address all market segments and both traditional and emerging workloads. I also believe Dell EMC (with AMD’s partnership) must do a few things to ensure success:
- Tell a compelling story: If the goal is to achieve a market leadership position, Dell EMC must tell a market leadership story—from the AMD partnership to enabling the enterprise, to drawing a distinction between Dell EMC and its competition. I believe the story is there, it’s just a matter of telling the market.
- Amplify the message: It’s easy to have success with the largest customers and pat yourself on the back. In order to lead in the market, however, Dell EMC has to take that compelling story and make sure the companies on Main Street “get it.” Small enterprise environments that must maximize every dollar of IT spend would love the Dell EMC – AMD story, but first, they have to hear it. Dell EMC would be wise to work with AMD in developing geographic specific campaigns that help amplify the partnership, the story, and the technology.
- Make the investments: Tied to my above point, Dell EMC and AMD must be willing to make the necessary financial commitments on the positioning, messaging, and marketing efforts of this PowerEdge lineup. They’ve invested billions in designing great products. Now it’s time to spend a little to sell them.
- Listen to the market: Dell EMC has done a great job of designing three servers that appear to have a compelling value prop for key workloads. That being said, nobody really knows where these servers are going to find traction until they start—well, finding traction. Dell EMC would be wise to be especially attuned to the market in these first few quarters of availability. The market will tell Dell EMC where to position today and what to design for tomorrow.
Dell EMC has quietly designed and delivered the most complete AMD EPYC-based server portfolio of the major OEMs. I believe Dell EMC has been very smart in its choice of platforms and the positioning (of those platforms), but the real test starts now. Dell EMC’s commitment to success in the marketplace will be made evident by the marketing and selling strategies behind this very compelling portfolio.
It is indeed an exciting time in the server market. I look forward to seeing where the Dell EMC – AMD partnership goes and how this portfolio evolves.
Note: This blog contains contributions by Moor Insights & Strategy principal analyst and founder, Patrick Moorhead.