Posts Tagged ‘Center’

Facebook Seeks Free Love Among Data Center Giants

October 29th, 2011

Facebook is dead serious about building common standards for efficient hardware in the data center.

On Thursday, the social networking giant launched the Open Compute Foundation, an industry association that aims to reduce the cost and environmental impact of the computers that run great swaths of the Internet. The foundation formalizes an effort the social media giant began in April, when it started the Open Compute Project (OCP).

In creating the Open Compute Project, Facebook “open sourced” its designs for a new data center that underpins its massive social networking service, and the foundation hopes to encourage other Internet icons to do much the same. The idea is to publish everything from the layout of server motherboards to the design of the warehouse-like buildings that house and cool the servers.

Open Compute members include Facebook, Intel, Dell, ASUS, Red Hat, Mozilla, Rackspace, NTT Data, and Netflix. Member organizations include corporations that make servers, software, networking equipment and power and cooling equipment, and companies that build and operate web-scale data centers.

The foundation is holding talks with the Open Data Center Association (ODCA), the user group for big data center operators founded by Intel. The foundation will take input from the ODCA’s hardware specification working group, said Frank Frankovsky, Director of Hardware Design and Supply Chain at Facebook and the face of the Open Compute Foundation.

Open Compute’s mission fills an urgent need, said Jason Waxman, general manager of High-Density Computing in Intel’s Data Center Group. The growth rate of server deployments will double over the next five years, he said. It will take “the equivalent of 45 coal power plants…just to keep up with that growth.”

Amazon Web Services illustrates the server boom well. Amazon’s daily increase of server capacity equals its entire capacity in 2000, when it had been in business for five years and pulled in $2.76 billion, said James Hamilton, an Amazon vice president and resident data center guru. Servers are close to the bottom line for Amazon Web Services. “That weighing point between being wildly successful and helping customers and being a tax on the company and not helping customers — the only difference is in the infrastructure,” said Hamilton.

In a related announcement, Facebook said it plans to build a massive data center just outside the Arctic Circle in the Swedish town of Lulea. The location means plenty of cold air to cool the servers and abundant hydroelectric power, which will meet most of the facility’s energy needs.

Open Compute’s specifications will standardize server motherboards, server racks, power supplies and cooling systems. This is analogous to the disk drive industry’s standardization of sizes and connectors, which made things easier for the industry’s customers and ended up boosting sales, said Andy Bechtolsheim, Founder and Chief Development Officer at Arista Networks, Inc. Bechtolsheim, who cofounded Sun Microsystems and was one of the first angel investors to fund Google, is on Open Compute’s board of directors.

The foundation is modeling itself on the open software movement and has taken its organizational cues from the Apache Software Foundation’s structure and bylaws, said Frankovsky. Open Compute’s bylaws are publicly available.

There are thousands of developers in the open software movement, including individuals who contribute bits and pieces when they have time. And there are many businesses built on open software. The realm of data center hardware is a bit different. In so far as Open Compute is a democracy, it is a democracy of the few. “In hardware, you need a supply-chain, you need a factory, you need access to low-cost parts,” said Bechtolsheim. “So it does benefit the larger vendors, let’s be honest here. You wouldn’t want to start a new company trying to build open racks.”

Open Compute is driven by equipment users — data center operators like Facebook, said Bechtolsheim. Server vendors differentiate their products in part on ancillary components at the expense of interoperability, he said. This greatly frustrates the data center operators, he said. “Open does away with gratuitous differentiation.”

Although the foundation’s efforts are likely to influence server design in general, the specs are aimed at helping web-scale data centers. “It’s not necessarily the ideal solution for every random commercial end-user that wants to add one more rack to their data center,” said Bechtolsheim.

Suppliers will also benefit from standardization in shared development costs, said Bechtolsheim. “There are only so many smart people in the world,” he said. “If everyone wants to write their own vertical standard, you waste all this intellectual capacity.”

Google is not an member foundation, but at Thursday’s launch, it was never far away from the discussion. As the behemoth among data center behemoths, the company has used its resources to make efficient, custom servers and data centers a competitive advantage.

Facebook has also put resources into designing efficient data centers, and much of its recent work is the core of the Open Compute Foundation’s initial specifications. By pivoting and fostering an open-software-like movement, Facebook can take advantage of collective development and more narrowly focus its internal development.

Of course, anyone else can do the same. “Apple wants to build a big iCloud. Obviously they want to minimize their power consumption and cost,” Bechtolsheim said. “I’m pretty sure they will look at this,” he said. And “they couldn’t have looked at this until it became an open spec.”

Facebook’s gain is twofold: by participating in the foundation they can more effectively influence their suppliers to meet their needs, and they can wrap themselves in the “open” flag at a time when Google is increasingly perceived as the new Microsoft. There is some irony in this, given the closed nature of Facebook’s product.

Open Compute is still finding its feet. It’s figuring out how to determine incubation committee membership. The nine-member incubation committee will select project proposals to forward to the board of directors for approval, said Frankovsky. Bechtolsheim is the committee’s chair.

The foundation is also wrestling with the issue of branding. One challenge is to keep companies from putting the OCP logo on everything so the brand doesn’t become diluted and mean nothing, said Frankovsky.

Source:http://www.wired.com/wiredenterprise/2011/10/facebook-open_compute/

FRAG Center to close after 7 years of business

November 1st, 2010

The combination of improved computers and a weak economy has led to FRAG Center closing its doors in the coming weeks after almost seven years in East Lansing.

The computer repair and video gaming lab, 543 E. Grand River Ave., will close after Nov. 12 because of a low profit margin, owner Scott Reschke said. The center’s repair business and video gaming customers have been decreasing since 2008, he said.

“It was like a very steep cliff,” Reschke said. “You climb and climb and climb, but there was no plateau. It just all of a sudden dropped and stayed down.”

The store’s business was steady from its opening in 2003 to 2007, with the latter being the best year for business, he said.

“2006 was just fantastic,” Reschke said. “I took out a bunch of money and remodeled the store because we just had gamers coming out of our ears.”

Then the combination of many changes in technology, as well as a slumping economy, brought profits to a minimum, he said. Software in computers has improved and reduced the need for spyware and virus cleanup, and hardware is more durable and requires less frequent repair, Reschke said.

“I never thought it would be us, but it just came down to the point of there’s not enough money coming in,” Reschke said.

FRAG Center was a welcoming environment for “nerds” to get together and play games in the same room, said Chris Graham, a Lansing Community College student who began going to the center about five years ago. Playing at the center was much better than being home alone and playing online, he said.

“My favorite part about coming here is it’s a room full of people playing the same game,” Graham said.

Another factor in the center’s closing was the lack of East Lansing’s free parking, Reschke said. Many people weren’t willing to pay to park and pay to use the computers, he said.

The FRAG Center was a valuable resource in the area for computer repair, said Julie Gold, a doctoral student at MSU. The closing of the store came as a surprise, she said.

“I’ve used them for years and they seemed to be the only place that could fix my computer and the service was always good,” Gold said.

Reschke said he now will focus on finishing his bachelor’s degree and raising his 2-year-old daughter. The best part of the FRAG Center was working with others and interacting with like-minded people, he said.

Source:-http://www.statenews.com/index.php/article/2010/10/frag_center_to_close_after_7_years_of_business

Taxation drives data center location

October 26th, 2010

The cost of power is a major consideration when locating a mega data center. But another factor is playing an increasing role: Taxes. “The site selection process has changed from low energy, to low energy and low taxes, to taxes and low energy,” says KC Mares, president and chief energy officer for Megawatt Consulting, which designs energy efficient data centers.

Income taxes are a major cost factor, and sales taxes are an especially big deal for companies like Amazon.com. “Only three states in US now have no state income taxes [and there are] very few with low sales tax that include hardware and software,” says Mares.

But the impact can be huge. Companies can save hundreds of millions a year in taxes simply by, for example, locating that data center in Nevada instead of California.

Source:http://blogs.computerworld.com/17231/taxation_drives_data_center_location

Data center consolidation and cloud computing in indonesia

June 27th, 2010

2010 brings great opportunities and challenges to IT organizations in Indonesia. Technology refresh, aggressive development of telecom and Internet infrastructure, with aggressive deployment of “eEverything” is shaking the ICT industry. Even the most steadfast division-level IT managers are beginning to recognize the futility in trying to maintain their own closet “data

irtualization, cloud computing, and drive to increase both data center economics and data security.

Of course there are very good models on the street for data center consolidation, particularly on government levels. In the United States, the National Association of State Chief Information Officers (NASCIO) lists data center consolidation as the second highest priority, immediately after getting better control over managing budget and operational cost.

In March the Australian government announced a (AUD) $1 billion data center consolidation plan, with standardization, solution sharing, and developing opportunities to benefit from “new technology, processes or policy.”

Minister for Finance and Deregulation Lindsay Tanner noted Australia currently has many inefficient data centers, very suitable candidates for consolidation and refresh. The problem of scattered or unstructured data management is “spread across Australia, (with data) located in not just large enterprise data centres, but also in cupboards, converted offices, computer and server rooms, and in commercial and insourced data centers,” said Tanner.

“These are primarily older data centres that are reaching the limits of their electricity supply and floor space. With government demand for data center ICT equipment rising by more than 30 per cent each year, it was clear that we needed to reassess how the government handled its data center activities.”

The UK government also recently published ICT guidance related to data center consolidation, with a plan to cut government operated data center from 130 to around 10~12 facilities. The guidance includes the statement “Over the next three-to-five years, approximately 10-12 highly resilient strategic data centers for the public sector will be established to a high common standard. This will then enable the consolidation of existing public data centers into highly secure and resilient facilities, managed by expert suppliers.

Source:-http://www.sys-con.com/node/1446330

Dell updates data centre solutions

June 13th, 2010

Dell has announced a new release of storage, networking and server hardware, plus services and systems management solutions focused on data centres.

The new product offerings are intended to enable businesses to add open-standard systems to extend existing data centres or for new deployments, with pre-configured solutions allowing companies to easily deploy virtualization solutions. The launch includes new Dell EqualLogic storage products, Dell PowerConnect networking solutions and Dell PowerEdge blade servers, new systems management updates and new storage, virtualization and support services from Dell.

“Organisations benefit from Dell’s approach to open, capable and affordable technologies that preserve choice and free up budgets for innovation and today we are introducing several new product and solution offerings to reinforce that direction,” said Brad Anderson, senior vice president, Enterprise Product Group, Dell.

With new solutions for virtualization management, new innovations in our server and storage product lines and expanded service offerings our customers can utilise Dell for the benefits of a converged architecture – dynamic workload allocation, faster time to deployment, seamless management – but they don’t want to be locked into a closed technology stack,” he added.

The new solutions have been designed for quicker deployment, with less resources and risk involved in rolling out solutions, along with integrated management solutions across, servers, storage and networking, to reduce the amount of resources involved in maintaining and managing solutions.

Among the new launches are Dell EqualLogic and PowerVault storage platforms, with automated data tiering, improved performance and scalability, and PowerVault systems for SMBs; and blade and rack PowerEdge servers, optimized for virtualisation, and updates to the LifeCycle Controller, Chassis Management Controller, and Integrated Dell Remote Access Controller (iDRAC) systems management solutions. The launch also includes new networking solutions, including PowerConnect-J series switches, gateways and chassis, produced through Dell’s OEM agreement with Juniper, and PowerConnect B-series chassis switches, produced through an OEM agreement with Brocade.

Source:-http://www.itp.net/580684-dell-updates-data-centre-solutions

Dell offers ‘open’ data center convergence

June 9th, 2010

Dell announced Wednesday “business-ready,” pre-integrated servers and storage arrays designed to host highly virtualized environments. The engineering represents Dell’s “open, modular, and pragmatic” response to the challenge of providing a “converged” data center.
PowerEdge blades and rackmount servers, based on the latest Intel and AMD processors, can be equipped with Infiniband, Ethernet, iSCSI or FibreChannel I/O devices that can handle existing one-gigabit Ethernet network speeds, more advanced 10-gigabit Ethernet or 40-gigabit FibreChannel speeds, Dell said.

The PowerEdge M610x and M710HD can be equipped with 192 GBs of memory for a greater density of virtual machines per host. The 610 and 710 models both can use up to 12 cores in two CPU sockets running either Intel Xeon 5520s or 5600s chips.
In a highly virtualized environment, servers are heavily loaded with memory and CPUs to handle multiple virtual machines. But they tend to become bogged down in I/O as the virtual machines require data off the network or send data to storage. By converging traffic, the I/0 devices — network interface cards and host bus adapters — don’t need to be divided for two different purposes. If the network is idle, then storage can momentarily make use of as many I/O devices as it needs to complete a burst to the storage array, and vice versa.

Virtualization is an enabling technology that allows a new level of flexibility and automation in the data center. Dell is focused on bringing the full benefits of that” to its customers, said Matt Baker, enterprise strategist, in an interview prior to Wednesday’s announcement.

In part, the offering is Dell’s answer to Cisco Systems’ converged networking and server environment, its Uniform Computing System with which it entered the server market in March last year. Dell’s offering is also a more concerted response to HP’s BladeSystem Matrix, which uses HP blades and switches to accomplish a similar goal.

For heavily virtualized servers to function at full potential, they need to be attached to a shared storage system that is being managed as a pool with a single file system.

Dell is addressing that need with software that manages new EqualLogic storage arrays that are half SAS disk drives and half solid state drives in a single enclosure, the PS6000XVS and the PS6010XVS.

With high-speed storage systems underlying their servers, Dell customers can use VMotion to maneuver virtual machines from one spot to another without interrupting the users using them, Payne said.

EqualLogic 5.0 firmware supports VMware’s vStorage API. Dell EqualLogic Host Software, such as EqualLogic Auto-Snapshot Manager, combined with the firmware can reduce SAN network traffic involved in a copy process by 95% and CPU usage by 75%, said Brian Payne, Dell product planning, in an interview.

The Dell storage and server configurations will work with Dell PowerConnect switches but “unlike competitors,” they will work with other vendors’ switches as well, said Baker.

Baker and Payne repeatedly emphasized that Dell is trying to give customers the benefits of a converged data center infrastructure, while sticking to open standards and known technologies to do so.

They said the benefits of converged server and storage management can be obtained with the pieces and architecture that Dell now supplies through its expertise in the x86 hardware field.

Some of that expertise comes from its Data Solutions business unit, which has been supplying servers to Microsoft, Ask.com, and Amazon for their Internet-based businesses. Microsoft uses Dell servers in its Azure cloud operations in Chicago.

Cisco uses a new standard approved by the IEEE that it helped write, an implementation of FibreChannel Over Ethernet known as Data Center Ethernet, in its Unified Computing System.

“The iSCSI protocol isn’t less efficient or slower than DCE,” said Payne. It’s more commonly implemented in one gigabit Ethernet settings because it works in what is a common, low cost networking environment today there. But it can ramp up to 10 gigabit Ethernet or 40 gigabit speeds if its users choose to.

He suggested Dell was protecting customers interest by highlighting a technology already well known and available as opposed to offering one that tied customers into particular product lines.

Source:-http://www.informationweek.com/news/software/server_virtualization/showArticle.jhtml?articleID=225600160&cid=RSSfeed_IWK_News

IT giants the greenest data center is the one that isn’t built

May 10th, 2010

Last month, Dell made the somewhat shocking announcement that it may never build another data center. The company was referring to the fact that it’s doubled its workload using no extra power and building no new data centers, simply by squeezing more capacity out of its existing servers. With an industry standard for data server utilization at about 12 to18 percent, there is ample room for improvement. What Dell realized was that by getting rid of its underutilized assets and swapping out the oldest and most outdated 25 percent of servers each year for the newest virtualization models, it would easily recoup its capital expenditures through reduced energy costs.

This might not sound revolutionary, but keeping legacy servers in place is a widespread habit in the IT world for much the same reason that people tend to keep their old refrigerator plugged into the garage or basement, when they have a new efficient one in the kitchen: they might be able to use it again someday. However, given how outdated both technologies are by the time they’re replaced, keeping them around does little but drain energy and cash.

At the Heartland GreenUp IT Conference in Des Moines last week Wells Fargo’s Executive VP of Enterprise Hosting Services, Jim Borendane echoed HP’s strategy, saying that the bank hopes to defer or avoid building new data centers. To achieve this goal, Wells Fargo has an aggressive “service life extension program” in which hundreds of servers that are under-utilized or more than ten years old are prioritized for decommission. As part of this process, existing space is renovated to current design techniques, such as keeping hot and cold air separate, and new technology, such as updated network switches, and servers capable of virtualization.

Also at the GreenUp, HP’s CIO, Randy Mott, explained how similar measures brought down the the cost of IT from four percent of HP’s revenue to the less than two percent that it represents now. Over a billion dollars were saved in 2009 by consolidating 85 data centers in 29 countries to just six data centers in three locations. Moss explained that because the best servers today are 19x more efficient than those from as recent as 2005, simply replacing them can help relieve two of HP’s biggest IT problems: sprawling server infrastructure and increasing costs.

While the greenest data center may be the one that isn’t built, sometimes new facilities will be required. Looking beyond server hardware, building design greatly impacts how energy-hungry a data center is. HP’snext generation data center in Wynyard, UK, which opened in February 2010, makes use of the wind coming off the North Sea not only to avoid the cooling costs for which data centers are notorious, but also to provide ten percent of the facility’s power from turbines. While a typical data center has a power usage efficiency (PUE) of 2.0 (meaning that it takes as much energy to cool as it does to run the equipment.

Last August, ACT’s data center in Iowa City, IA, became the first data center to receive a LEED Platinum certification from the USGBC. Also represented at the GreenUp, ACT’s assistant VP Tom Struve and Kevin Monson of Neumann Monson Architects explained how incorporating sustainable design elements actually improved the data center’s ability to perform its primary function and aid in the building’s resiliency to Iowa’s often extreme weather conditions. The LEED design guidelines prompted ACT to develop a custom monitoring system that trends key electrical and cooling parameters, compares usage to installed capacities and allows ACT to proactively predict and plan for modular capacity expansion.

Source:-http://www.triplepundit.com/2010/05/green-data-center-dell-greenup-it/

Get Adobe Flash playerPlugin by wpburn.com wordpress themes