Accelerating Network Computing Performance with All New Architecture

Background

With the exponential growth of mobile Internet devices and the rise of Cloud Computing, end-users demand faster, more powerful and more secured Internet services. This unprecedented demand has presented a new challenge which Internet service provider and IT system operators maintain and manage their network traffic. To ensure smooth network traffic, an all new enhanced architecture is necessary in order to handle the various types of data packets and security instructions. In other words, performance, power efficiency, scalability and reliability are the major factors in network traffic management.

Accelerated Performance with All New Architecture

To meet the challenges of the multiplying data packet delivery and increased security concerns, one practical solution is the adoption of dual Intel® Haswell-EP Xeon® E5-2600 v3 CPUs, combined with Wellsburg PCH. The all new enhanced, scalable and easily manageable architecture, codenamed “Grantley”, is announced to be the next step of Intel® platform roadmap with a determination to reinforce performance in network computing.

Major Advancements

One of the most significant benefits is the breaking-through acceleration of power/performance efficiency. The new Xeon® E5-2600 v3 CPU supports Per Core P-state (Performance State), programmed to control the clock frequency and voltage of each independent core. It is announced by Intel® to improve energy efficiency by up to 36%, which also greatly reduces the ownership cost for network facility administrators.

Like its former v2 CPU platform (codenamed “Romley”), the Intel® Xeon® E5-2600 v3 processor also offers up to 16 cores and dual CPU design, with the aim of performance acceleration. Aside from multiple cores and processors, the new platform determines to further strengthen Intel® QPI (QuickPath Interconnect) up to 9.6 GT/s. This will give an even greater lift to inter-CPU communications and data flow, compared to its former v2 generation.

Regarding the system memory, Grantley is the first network platform supporting DDR4 memories, with up to 2133 MHz clock frequency and lowered operating voltage to 1.2V from 1.5/1.35V. This is a great step for caching power, performance upgrade and greener structure from previous DDR3 implementations. Based on statistics from Intel®, this will improve performance and power efficiency by nearly 50%.

Applications in Virtualization

Like its previous Xeon® generations, the Intel® Haswell-EP Xeon® E5-2600 v3 CPU features Intel® Virtualization Technology (Intel® VT), a hardware-assisted virtualization tool for software-based virtualization solutions. This upgrading mechanism not only reduces virtualization overhead of CPU utilization, but also improves data throughput, flexibility and reliability through hardware assistance. With hardware support, the involvement of virtual machine monitoring for I/O traffic can be reduced in order to gain higher virtualization efficiency.

On the other hand, by taking VM (Virtual Machine) service quality into considerations, Intel® implements Cache Monitor (Cache QoS) in its Grantley platform for the first time. Without Cache QoS, cache works on a FIFO (First-In-First-Out) basis. In conventional VM applications, larger VMs and smaller VMs both utilize the FIFO approach. This would downgrade the virtualization performance in a long run. On the other hand, the all new mechanism monitors and provides information on individual VMs in cache and makes better utilization of cache spaces. This will greatly improve virtualization efficiency in long terms.

Lanner and Intel® Grantley

With the adoption of the latest Intel® Grantley platform, Lanner FW-8896 leverages all the technological advantages from above. Aside from performance and efficiency upgrades, FW-8896 also delivers high scalability.

The Intel® Grantley based networking appliance offers the highest LAN port density. FW-8896 supports a wide range of NIC modules, up to 8 Ethernet modules with 64 x 1GbE, 32 x 10 GbE, or 16 x 40 GbE ports for customization, a much wider bandwidth than other similar devices in the market.

Regarding Lanner’s NIC Ethernet expansion options, the latest NCS2-IXM407 and NCS2-IQM201 are both built with the latest Intel Fortville 40GbE Ethernet controller supporting up to 4 x 10GbE and 2 x 40GbE broad bands respectively. Interface wise, NCS2-IXM407 is built with SFP ports and NCS2-IQM201 is designed in QSFP ports, more panel-efficient than conventional LAN ports. The NIC Ethernet modules can be complemented by PCIe Gen 3.0 lanes.

To accelerate cryptographic and data compression performance, FW-8896 can be configured to implement Intel® Coleto Creek chipset which comes with built-in Intel® QuickAssist Technology to optimize virtualization and communication applications. This option would highly boost the processing rates of network traffic.

Other reliability features include N+1 hardware cooling fans and 1+1 ATX 800W redundant power supply designed to run operations at a constant base.

Accelerating Network Computing Performance with All New Architecture was last modified: January 14th, 2020 by LEI Technology