While its large portfolio of SKUs for any and all use cases has always been a key part of Supermicro’s value proposition to the channel, they are emphasizing their Resource Saving Architecture’s power and saving capabilities in the refreshed models.
SAN FRANCISCO – Today, Intel unveiled its new 2nd Generation Intel Xeon Scalable [Cascade Lake] processors at a launch here, and OEM partner Supermicro joined in the festivities, unveiling its Cascade Lake-enabled refresh of its entire portfolio of X11 servers, storage systems and Server Building Block Solutions.
“We are incredibly excited about this launch,” said Peter Han, Supermicro’s Chief Marketing Officer. “We have always had a strength in fast silicon transitions and getting products out before our competitors. We have offerings for over 170 different server configurations with workloads optimized for the Cascade Lake processors. All of them have been transitioned to the new silicon.”
A key differentiation for Supermicro in the server market is the vast array of specialized configurations it makes available to cover a large variety of specific workloads and use cases.
“We have optimized these for all sorts of workload types and for considerations like performance, density, and cost effectiveness,” said Jim Sangster, Senior Director of Product Marketing. “That allows a partner to match to a specific workload, size and form factor, whether a rackmount 1U or embedded. With those parameters in mind, we have a no compromise workload optimized design for whatever you need – which goes back to why we have no many SKUs.”
The core performance advantages of the new 2nd Gen Intel Xeon Scalable processors include the Intel Optane DC persistent memory to increase memory capacity and affordability, and Intel Deep Learning Boost to enable more efficient AI acceleration.
“Intel has been working with partners for years on bringing Optane to market, and from our initial testing, the memory management within Cascade Lake has unleashed all kind of innovation,” Han said. “The impact of the memory in PCs was more limited, but with Cascade Lake, those years of development are paying off. The application recovery time has been reduced from minutes to seconds – 12.5x for SAP HANA. There is an 8x improvement for Analytics with Apache Spark SQL database.”
“The memory footprint, at 1.5x memory per socket, is larger than ever before,” Sangster said. “It gets well past the 3 TB per socket limit that existed previous. It’s also cheaper, so a mix of DDR4 and Optane is cheaper than just DDR4.” The processors also have 10 per cent faster DIMMs, and faster CPU frequency, up to 3.8GHz.” It all translates into what Supermicro says will be up to 35 per cent better data centre performance, and up to 50 per cent better TCO.
Supermicro is also emphasizing the green-ness of its new offerings.
“At Supermicro, we are also committed to minimizing the environmental impact of our server technology, with our unique Resource Saving Architecture,” Han indicated. “TCO is very well understood, but we also emphasize TCE [Total Cost to the Environment].” This architecture disaggregates the CPU and memory along with other subsystems, so each resource can be refreshed independently, allowing data centres to reduce refresh cycle costs and reduce e-waste. Further savings are achieved through shared power and cooling as well as free-air cooling solutions.
“Our 2U BigTwin and 4U FatTwin families use this architecture to provide shared power and shared cooling which are very efficient,” Sangster said. “Our SuperBlade family also has this disaggregated design.” The SuperBlade systems come with two-socket and four-socket Intel Xeon Scalable processor-based blade servers supporting top-bin 205-watt processors, NVMe, 100G EDR InfiniBand switch, or 25G/10G Ethernet switches, redundant AC/DC power supplies, and Battery Backup.
Supermicro is also launching a range of their high-end 1U and 2U Ultra SuperServers, which feature dual Cascade Lake processors [up to 28 cores, 205W TDP per CPU], 24 DIMMs of DDR4-2933MHz Reg. ECC memory, up to 24 hot-swap NVMe [up to 16 million IOPS] or SAS3/SATA3 drives, up to eight PCI-E 3.0 expansion slots, dual or quad-port 1G, 10GBase-T, 10G SFP+, or 25G SFP28 Ethernet, and redundant 750W/1000W/1600W Titanium Level (96%+) digital power supplies.
The new Cascade Lake Petascale line of all-flash NVMe 1U storage servers, a broad selection of servers optimized for AI, Deep Learning, and HPC workloads, and new multi-processor 4-socket and 8-socket servers idea for large memory applications like in-memory databases and real-time analytics were also announced. So were several Intel Select Solutions for Supermicro, verified hardware and software stacks optimized for specific workloads across compute, storage, and network. Supermicro also announced that it anticipates new solutions on the 2nd Gen Intel Xeon Scalable processor, including Intel Select Solution for VMware vSAN V2, Intel Select Solution for SAP HANA TDI, and Intel Select Solution for Hardened Security with Lockheed Martin
Many of these SKUs are available on Day One, but not all.
“It depends on what server and processor it is,” Sangster said. “We have a majority coming out on day one and some quickly thereafter. Some processors are Wave 2, Wave 3 and beyond.”
“Supermicro is always about speed to market, and that means making a lot of the SKUs available right out of the gate,” Han said.
Supermicro is also very friendly to the channel,” Han concluded. “We grew up here and we remember our roots.”