At the Blade.org gathering, technologists from Morgan Stanley and The Friday Institute for Educational Innovation at North Carolina State University and GameVee.com joined with leading suppliers of blade-based data center gear such as APC, BLADE Network Technologies, Brocade, Devon IT, Emulex, IBM, Intel, NetApp, NetXen, QLogic, VirtenSys and VMware. These visionaries detailed the emerging mega trends of converged networks, advanced energy efficiency and hyper consolidation, which will unify data center I/O on a single wire, reduce energy waste and enable unprecedented data center densities.

Since its inception two years ago, more than 200 companies have joined Blade.org to advance open innovation around blade server-based data centers.  The newest members include MetLife, Morgan Stanley and CBS Television.  The consortium’s membership also includes leading end-users from financial services, media and entertainment and academia together with top IT vendors of data center hardware and software, services and solutions.

The Mega Trends that Blade.org believes will change the future of business computing are:

Converged Networks: Today data centers use separate “input/output (I/O) lanes” for data networking, storage traffic and inter-process communications, each of which require their own adapters, connectors and wires. In the future, data centers will converge I/O on Ethernet, running all their traffic on a single "lane" or wire. This convergence is ideal for blade environments because they are tightly packed together with little room for extra components.

“Data centers can begin their move to single-wire fabrics that transport Fibre Channel storage data over lossless 10 Gigabit Ethernet networks, which enables the convergence of storage, networking and clustering advised Symposium presenter Tim Chao, VP, Software Engineering and Advanced Technology for BLADE Network Technologies. “Data centers that deploy Fibre Channel ofver Ehternet (FCoE) as part of a unified data center I/O strategy can realize key benefits such as increased performance, greater data center density, lower capital and maintenance costs and lower energy and cooling costs.”

Advanced Energy Efficiency: Use of blade technology is anticipated to represent more than 25 percent of server deployments by 2010. The need to dissipate the heat generated by this highly power dense technology must keep pace; however, in recent years, traditional cooling system design has proven inadequate to remove concentrated heat loads. In the future, data centers will minimize energy use and reduce carbon footprints by turning to such alternatives as water-cooled and/or container-based data centers and close-to-chassis heat rejection that can eliminate much of the 45 percent of electrical energy in the data center that is wasted by today’s cooling systems. Energy efficiency was a big part of the discussion at the Symposium.

"Blade servers are part of the solution because they are much more energy efficient than a rack full of 1U servers," he says. "But at the same time, because of the ability to provide a high-density offering within a smaller area, they're consuming more energy in a smaller space,” advised Doug Balog, VP, Blades and Modular Development, IBM Systems & Technology Group and Chairman of Blade.org.

“Gartner states that in-rack and in-row cooling will emerge as the predominant cooling strategy by 2011 for high-density equipment,” said Blade.org Technology Symposium presenter Jim Simonelli, chief technology officer of APC.  “This strategy closely couples the cooling unit to the heat emanating from the rack or row of servers.  The design eliminates performance uncertainties and hot spots that exist -- especially when data center professionals attempt to cool high-density racks using a perimeter cooling strategy.  Row-oriented cooling also improves energy efficiency since this approach reduces the distance cold air must travel to cool the servers, resulting in reduced fan power -- the dominant source of electrical energy consumption in a data center's cooling system.”

“There will be a real focus at the server ingredient level to reduce power requirements and improve energy efficiency while increasing performance,” said Symposium presenter Rob Nance, Chief Architect, High Density Computing Group for Intel. “The result will be that data center managers can deploy more compute power in their existing facilities without upgrading their infrastructures.”

Hyper consolidation: Today data centers are focused on server consolidation. In the future, clients will “hyper consolidate” servers, workstations and network devices into an integrated blade environment. For example, Blade.org member Devon IT’s new approach to dense desktop consolidation on blades achieves exceptional scalability, improved security and tremendous power savings by consolidating desktops on blade servers and delivering graphics and multimedia to the user directly from blade installations.

"Blades in the data center will evolve to a hyper-consolidated model through virtual appliances in which all discrete servers, firewalls and other network devices will be consolidated into the blade chassis," VMware chief platform architect Richard Brunner advised the Symposium audience.

Blade.org’s Unique Collaboration Culture
In a new academic paper from the Pennsylvania State University Department of Management and Organization called “Blade.org: A Collaborative Community of Firms”, Professor Charles C. Snow writes, “Founded in early 2006, Blade.org, a successful community of member firms, was creatively designed to emulate the characteristics and processes of a community of individuals. We believe that the community-of-firms model is particularly well suited to the pursuit of continuous innovation, and we expect to see this organizational approach used increasingly as firms come to realize the value to be gained from multi-firm collaboration both within and across industries. Blade.org, the pioneer of this new approach, shows that a community of firms can be a successful means of innovation, and its role as a community facilitator is worthy of study and imitation.”

At the Symposium, IBM’s Dr. Tom Bradicich, IBM Fellow, and Vice President, Systems Technology, IBM Rack, Blade, and x86 Servers and Chairman of the Blade.org Technology Committee, highlighted some examples of this collaboration as:

Green Data Center Monitoring – An APC power management, IBM BladeCenter and TAC Building Management solution integrates power and thermal metrics with IT to provide management tools that are energy-aware and can assess the ;/;/impact of energy on business services.
Power Savings for SMB Data Centers – A new Blade.org solutions reference architecture is targeted specifically at power savings in small and medium-sized business (SMB) environments. This reference architecture using component elements from APC, IBM and Kell Systems outlines suggested power saving configurations and best practices for small data centers
Fibre Channel over Ethernet (FCoE) Demonstration – BLADE Network Technologies, Emulex, NetApp and QLogic have collaborated to demonstrate a blade server-based Fibre Channel over Ethernet (FCoE) unified fabric. The demonstration shows that the “loss-less” I/O required to carry Fibre Channel over Ethernet (FCoE) storage traffic within blade architectures is now possible using products based on the emerging standards for Converged Enhanced Ethernet™ (CEE), an enhanced version of Ethernet for data centers.

Reuters Market Data System Performance Test – An IBM BladeCenter H equipped with Red Hat Enterprise Linux and 10 Gigabit Ethernet switching from Blade Network Technologies and Chelsio Communications has achieved the best performance for demanding workloads in the financial services industry. A recent test conducted by Securities Technology Analysis Center (STAC) revealed the lowest mean latency and the lowest standard deviation of latency ever reported in achieving record-breaking performance for the Reuters Market Data System (RMDS).

Blade servers have long been signaled as the wave of the future, but Steve Russell, managing director and global head of enterprise computing at Morgan Stanley, advised the Symposium audience that the future of blades is now.  “We first started using blades in 2002 and we were talking about how many chassis we wanted,” said Russell.  “Then last year we were talking about how many racks we wanted. Now it’s, how many thousands of systems do I want delivered in a given moment?”

Comments (0)

Subscribe to this comment's feed