Data Center Journal

VOLUME 57 | OCTOBER 2018

Issue link: https://cp.revolio.com/i/1036193

Contents of this Issue

Navigation

Page 17 of 20

In today's hyperscale data center, the focus has moved from designing servers to designing racks, giving us the aptly named rack-scale architecture. The Open19 Foundation was designed to give the data center industry a hardware platform around this type of architecture and to create standards and designs that are customizable, flexible and economical—and that fit any 19" rack environment. (Hence the name "Open19.") BY MD TRUONG T hrough collaborations among global industry-technology leaders, Open19 solutions focus on operational simplic- ity, power efficiency and flexible de- signs—all important considerations for rack-scale architectures. And with the focus shi- ing from a server-centric world to a rack-centric world, data center managers can break conven- tions and solve a common challenge: operations. Hyperscale data centers are vast operations, oen spanning an area the size of multiple football fields. Behind hardware acquisition costs, the operational expenses of running hardware at scale affect the total cost of ownership (TCO). Power-supply efficiencies (or, more commonly, inefficiences), airflow obstructions and simple hardware servicing add up quickly with a foot- print that large. To compound the problem, these sites are serviced by only a handful of on-site technicians, meaning repairs and upgrades oen take time. Open19 was developed to address that opera- tional challenge at scale. e founders of the col- laboration—which includes nearly 30 companies, including Molex—recognized that the cost of rack integration, operation and maintenance was be- coming a financial burden, and their original goal was simple: reduce integration costs, including time spent placing hardware in the rack, cabling servers and networking, and dressing cables for airflow. rough collaboration with other leading technology companies, Molex worked to develop a solution that eliminates those high integration costs: the external cableized backplane, capable of delivering up to 100 Gbps Ethernet connectivity to each server brick, or one half-wide 1U-tall server. e cableized backplane solution is based on the same technology in backplane connec- tors, which typically serve in internal blind-mate applications. is core technology was enhanced to accommodate the new Open19 standard, and it means users can discard the high-touch QSFP, QSFP-to-SFP breakout DAC cables and Cat 5 Ethernet cables typically used for Ethernet. e cabled backplane brings improved airflow and simpler installation thanks to fewer installation errors and better signal integrity for high-speed Ethernet. e cabled backplane resides in a shell to create a spine backbone for 12 bricks. THE DATA CENTER JOURNAL | 15 www.datacenterjournal.com

Articles in this issue

Links on this page

Archives of this issue

view archives of Data Center Journal - VOLUME 57 | OCTOBER 2018