The 10 Top Enterprise Servers Of 2018 (So Far)

Coolest Server Tech In 2018

Worldwide enterprise server revenue for the first quarter of 2018 reached an all-time high of $18.8 billion as sales increased a whopping 39 percent year over year.

High server demand this year is being driven by enterprises and SMBs seeking refreshes, new technology to build software-defined infrastructure, and more deployments of next-generation workloads. This year has already seen Dell EMC and Inspur launch new servers to support high-density workloads and accelerate application performance, while IMB's new Power System AC922 was built to take on the world's biggest artificial intelligence challenges.

CRN breaks down the enterprise server technology in the market in 2018 so far.

(For more on the biggest news of 2018, check out "CRN's Tech Midyear In Review.").

Cisco UCS C4200

Cisco engineers spent a year creating the new Unified Computing System C4200 Series Rack Server Chassis and nodes targeting compute-intensive, scale-out workloads. Cisco's new C125 M5 Rack Server Node has up to 128 percent higher processor core density compared with a 42RU standard rack of UCS servers. The C125 contains two AMD Epyc 7000 CPU processors. All previous Cisco UCS products have used Intel processors such as Xeon. The nodes are part of Cisco's new C4200 chassis, which is a high-density, multi-node solution that maximizes application performance and operational efficiency. Each node contains 16 DIMM slots for up to 2 TB of capacity and 2666MHz of DDR4 memory.

Dell PowerEdge R940xa

Dell EMC launched its new four-socket PowerEdge R940xa server at Dell Technologies World in May. The server is purpose-built to tackle different performance bottlenecks with improved response times and reduced statistical errors compared with older PowerEdge generations. The R940xa was designed to accelerate applications to deliver real-time decisions for GPU database workloads. It combines up to four CPUs with four GPUs in a 1:1 ratio and enables low latency with direct-attached non-volatile memory express (NVMe) drives. The product is optimized for workloads that are compute-layer-heavy and take advantage of CPU RAM to move data into GPU RAM for faster computation.

Dell PowerEdge R840

Dell also launched this year its new four-socket PowerEdge R840 server. The R840 is designed to "turbocharge" data analytics," said Ravi Pendekanti, senior vice president of product management and marketing for Dell's Server and Infrastructure Systems, when the R840 was unveiled. The server speeds up response times with up to 24 direct-attached NVMe drives, while also accelerating critical workloads with up to two GPUs or two FPGAs. "We're able to support up to 24 direct-attach NVMe drives, which we believe is about two times above the nearest competitors we see," Pendekanti said.

HPE's Superdome Flex

Hewlett Packard Enterprise's new four-socket Superdome Flex server is based on Intel's Skylake Xeon processors. The Superdome Flex product marks the first time that HPE has combined the robust SGI high-performance modular compute NUMAlink technology, which it acquired from SGI in 2016 for $275 million, with the breakthrough price performance of the x86 Intel Xeon platform. The server offers scalability of up to 32 sockets, in-memory design and memory capacity of up to 48 TB.

Huawei KunLun V5

Huawei unveiled its KunLun V5 x86 mission-critical servers with technology partner SUSE powered by Intel Xeon Platinum 8100 and Gold 6100 processors. The eight-socket KunLun V5 was designed for core databases and mission-critical applications in a wide range of industries with up to 32 CPUs in a rack. With RAS 2.0 technology, the server offers an open x86 ecosystem with the reliability of Unix servers. There are three models to the KunLun V5: the 9008 with one eight-CPU chassis, the 9016 with two, and the 9032 with four chassis.

IBM Power System AC922

IBM's new Power System Accelerated Compute Server, the AC922, was designed for high-performance computing, analytics and artificial intelligence, enabling enterprises to easily deploy data-intensive workloads. The AC922 contains Power9 processors that supports up to 5.6X more I/O and 2X more threads than its x86 competitors, according to IBM. The Power9 is available on configurations with anywhere between 16 to 44 cores in the AC922. The server also contains a variety of next-generation I/O architectures including PCIe Gen4, CAPI 2.0, OpenCAPI and Nvidia's NVlink technology. The company said the server was built for the world's biggest AI challenges.

Inspur i48

Inspur launched its new i48 Server in June optimized for data center application scenarios, supporting up to eight high-density compute modes, four computing-storage balanced nodes and two large-capacity storages nodes in a 4U chassis. The product is aimed at supporting the infrastructure requirements for artificial intelligence, HPC and network function virtualization (NFV). The i48 supports 100G of high-speed connection of all nodes to core switch with low-latency, or two 48-port 100G TOR switches to form a large-scale system via cascade in single rack.

Inspur Yingxin NF5486M5

Inspur's new high-density Yingxin NF5486M5 server is designed for large-scale data centers such as cloud service providers and telecom operators. The 4U server is optimal for application scenarios of warm and cold data including video storage, archival storage and application scenarios in big data and HPC. The Yingxin NF5486M5 has storage capacity of 1.25 PB and supports 106 hard disk drives. With an on/off control module configured for each hard disk drive, the server adjusts real-time power according to the workload of each disk. The server also includes smart dynamic control technology for power supply and thermal dissipation.

Lenovo ThinkSystem SR950

Lenovo's ThinkSystem SR950 is designed for demanding, mission-critical workloads such as in-memory databases, large transactional databases, batch and real-time analytics, and virtualized server workloads. The 4U server can scale from two to eight Intel Xeon Scalable processors with up to 12 TB of memory in 96 slots. The server's modular design provides rapid upgrade and servicing with easy front or rear access to all major subsystems, helping to reduce service costs. Operating systems supported in the SR950 including Microsoft Windows Server, Red Hat Enterprise Linux, VMware vSphere and SUSE Linux Enterprise Server.

Supermicro SuperServer 4029GP-TRT

Through a technology partnership with Nvidia, Supermicro unveiled its SuperServer 4029GP-TRT this year, which supports eight Nvidia Tesla V100 32-GB GPU accelerators with maximum GPU-to-GPU bandwidth for cluster and hyper-scale applications. The 4029GP leverages Intel Xeon Scalable processors with up to 28 cores with Intel HT Technology. The system includes 24 DIMM slots and up to 3 TB of memory capacity. The server incorporates the latest Nvidia NVLink technology and features independent GPU and CPU thermal zones to ensure performance and stability under the most demanding workloads.