The 10 New Enterprise Servers Of 2019 (So Far)

New servers for 2019 are increasingly being configured with such technologies as high-end GPUs and high-performance NVMe flash storage in order to be ready for new performance-hungry applications such as artificial intelligence, machine learning and big data analytics.


While the overall number of servers shipped seems to be declining, the revenue of those servers continues to show growth, according to IDC's Worldwide Quarterly Server Tracker for the first quarter of 2019.

For server vendors, however, there is little slowing down in the adoption of new technology. New servers for 2019 are increasingly being configured with such technologies as Nvidia GPUs or high-performance NVMe flash storage in order to be ready for new performance-hungry applications such as artificial intelligence, machine learning, big data analytics and more. And those servers are increasingly being called upon to manage workloads such as IoT at the edge where processing of certain types of data can be done more efficiently than in central data centers.

CRN looks at 10 new enterprise servers aimed at new ways of computing.

Sponsored post

Amax BrainMax AI Servers

Custom server builder Amax in June introduced its new BrainMax RTX series of servers that come preintegrated with Nvidia Quadro RTX GPUs and CUDA-X software to provide artificial intelligence to solving big data analytics, machine and deep learning, and AI workloads. The Fremont, Calif.-based company's top-of-the-line BrainMax QR-480 is a 4U server configured with eight Quadro RTX 6000 or RTX 8000 GPUs, and is validated for on-demand rendering and virtualization capabilities. It can be connected to multiple high-end virtual workstations supported by Nvidia Quadro Virtual Data Center Workstation (Quadro vDWS) software for remote, shared and multi-GPU configurations.

Asus ESC4000 DHD G4

Asus, San Jose, Calif., in March introduced the ESC4000 DHD G4, a new 1U general- purpose GPU server for edge computing applications. The ESC4000 DHD G4 supports up to four Nvidia GPUs for AI and high-performance computing for heavy workloads including 3-D rendering. The new servers also include NVMe storage connectivity with four front-panel hot-swap bays.

Dell EMC DSS 8440

Dell Technologies in April used its Dell Technologies World to unveil the new Dell EMC DSS 8440, a two-socket, 4U server targeting machine-learning applications and other high-performance workloads. The Dell EMC DSS 8440 can be configured with four, eight or 10 Nvidia Tesla V100 GPUs, and includes high-speed I/O with extensive local storage for compute-intensive workloads such as modeling, simulation, and predictive analysis in scientific and engineering environments. An open architecture based on industry-standard PCIe allows customization of internal components such as accelerators, storage options and network cards.

Fujitsu Primequest 3800E2

The Primequest 3800E2 server introduced by Fujitsu in March is a 7U, eight-socket rack server targeting mission-critical environments. The server includes the latest Intel Xeon Platinum processors with up to 28 cores per processor and DDR4 memory modules, with memory capacity of up to 24 TB using only DDR4 or 36 TB using the new Intel Optane DC Persistent Memory. The Primequest 3800E2 helps improve performance for in-memory databases such as SAP HANA and Microsoft SQL Server 2017. Fujitsu said that, because of the server's reliability, availability and serviceability capabilities, it can be used for running Unix workloads and migrations.

Gigabyte G191-H44

Taiwan-based Gigabyte in May unveiled the G191-H44, a new server equipped with Nvidia EGX T4 GPUs to target edge computing and IoT requirements. The G191-H44 is a 1U server featuring all the capabilities of Gigabyte's G-series server family, including its airflow-optimized design and large thermal envelope. It can be configured with four Nvidia Tesla T4 accelerators for use as an edge server or as part of an edge micro data center, or with Nvidia Tesla V100 Tensor Core accelerators for use in centralized data center AI computing.

Inspur NF5488M5

China-based Inspur in March released its NF5488M5, calling it the industry’s first AI server to support eight Nvidia V100 Tensor Core GPUs interconnected with a high-bandwidth NVSwitch in a 4U form factor. The Inspur NF5488M5 is targeted at such deep-learning and high-performance computing applications as voice recognition, video analysis and intelligent customer service. The server's eight Nvidia Tesla V100 Tensor Core 32-GB CPUs provide 5,120 Tensor cores with up to 1 petaflop of AI. computing performance. Customers can add two 28-core CPUs for top-level, general-purpose computing performance, and 6 TB of persistent memory for high-speed data access. The server also includes flexible GPU cluster expansion over PCIe fabric.

Lenovo ThinkSystem SE350

Lenovo used its March presence at the MWC Barcelona conference to unveil its new ThinkSystem SE350, an edge server the Research Triangle Park, N.C.-based company said is just slightly larger than a mobile PC. The new purpose-built server is aimed at deployment in tight spaces, typically outside the data center.

Lenovo said the rugged SE350 can handle temperatures from 32 degrees to 131 degrees Farenheit, and tolerate locations with high dust and vibration. It features the Intel Xeon-D processor with up to 16 cores, along with 256 GB of RAM and 16 TB of internal solid-state storage. The SE350 also supports the Nvidia Tesla T4 for workloads such as edge inferencing for AI deployments.

Penguin Computing Relion XE2142eAP

Penguin Computing, the Fremont, Calif.-based specialist in Linux-based servers, in June introduced the Relion XE2142eAP, a new server based on Intel's new Xeon Platinum 9200 series processors. The Relion XE2142eAP offers up to 448 cores in a 2U, four-node, liquid-cooled configuration, or 224 cores in a 2U, two-node, air-cooled configuration. The servers also support two M.2 NVMe drives or two M.2 and two U.2 NVMe drives for high-density storage optimized for high-performance computing and AI workloads.

Supermicro SuperServer 2029BR-HER

Supermicro in June introduced what it called the first server to support EDSFF, or Enterprise and Data center Storage Form Factor, NVMe drives. The San Jose, Calif.-based company's SuperServer 2029BR-HER, part of its BigTwin family, is a 2U, four-node server with room for 10 E1.S EDSFF NVMe drives plus two SATA M.2 SSDs per node. Each hot-pluggable node also supports one or two second-generation Intel Xeon Scalable processors, 24 DIMMs with up to 6 TB of DDR4-2933MHz memory, two PCIe 3.0 slots, and IMPI (intelligent power management interface) with KVM and dedicated LAN.

Tyan Thunder SX TN76-B7102

Taiwan-based Tyan in May introduced a new version of it Tyan Thunder SX TN76-B7102 edge server that integrates the Nvidia EGX platform to help overcome bottlenecks in cloud computing-based artificial intelligence applications. It is a 2U, dual-socket second-generation Intel Xeon Scalable processor-based system that includes eight PCI3 x8 slots or four PCIe x16 slots. It support up to four single-wide Nvidia T4 GPU cards, and has room for 12 3.5-inch SATA drives.