What is fundamentally wrong with data centers of the past is it was an exercise in consensus decision making to gain enough votes of confidence to move forward with a major capital investment. The Real Estate and IT group would go around to all the different business units and other parts of the company to collect the requirements, and alternatives would be presented. Here is how much it will cost to meet the needs of the business for the next 15 years.
This method was fine when Data Centers were a fraction of the IT costs. Now with the web and surge of data, it is reasonable for the top financials to have 50k - 100k of servers. Some of these servers need to meet regulatory requirements from dozens of gov't agencies. Some of these servers have minimal regulatory issues and can be spread around in low cost data centers. Google, Amazon, Facebook, and Microsoft are building some of the lowest cost data centers that are tuned to their high server counts with geo-redundant homogenous architectures. Google has 5 data centers support a major geographic region for ad services where 24x7x365 services are requirement. In contrast the gov't financial regulators will require an A + B data center and disaster recovery site strategy within a specific distance of Manhattan. These requirements push the costs of data centers to be one of the highest costing in the industry as Active-Active fail over with full capability to run services if the other goes down.
With todays current financial climate, it is time for a change.
And one of the first financials to make the change is Goldman Sachs. Don Duet is quoted in the press release.
"Their innovative technology and services will allow Goldman Sachs to scale its data center operations more efficiently, and further advance the firm's broader commitment to environmental stewardship and reduced carbon footprint."
The three points that GS focuses on are part of a green data center strategy.
1) Efficient operations
2) Commitment to environmental stewardship
3) Reduced Carbon footprint
The money savings isn't mentioned by Don in the quote, but it is highlighted as a feature of a data center 2.0 strategy
In addition to greater operating and capital expense savings
Add all these things up and one way to look at GS's strategy is to gain the capabilities that Google, Facebook, Amazon, and Microsoft have to design data centers that meet their business needs in a way that data centers capacity can be deployed yearly vs. the past where data centers were built every 3-5 years.
The WSJ highlighted the GS announcement as part of a post here.
Data- and technology-driven organizations like Goldman Sachs are particularly vulnerable to the pace of technological change, because the huge investments they make today could cripple them tomorrow. Whatever competitive advantage they may have earned today can be swept away in the next tide of change, particularly if their hardware investments prevent them from reacting in an agile manner. That’s why Don Duet, the global co-chief operating officer of Goldman’s technology division, is building modular data centers that depend more on software than hardware, so that his team can react to “the pace of technological change,” he said during a phone conversation Monday.
WSJ also reported on Allianz Global Investors questioning its data center strategy.
The economic crisis in Europe is forcing Allianz Global Investors of America to reconsider its data center consolidation strategy. Daniel Stroot, CIO of Allianz, says the company considered opening two data centers in Europe and two in Asia, in addition to the two it maintains in the U.S., but is now planning to add just one more. “We had planned to have two in each region but now we’re thinking maybe we only need three globally,” Stroot told CIO Journal in an interview. “The crisis in Europe has continued to force us to look at being more efficient.”
At this point, the company no longer owns its own data centers, having in 2011 consolidated five data centers down to two private clouds operated by data center service provider IO. The original reasons for consolidation were as much circumstantial as a reaction to other changes in technology, namely software-as-a-service. The company was moving two of its offices into new buildings in New York and San Francisco that either wouldn’t support a data center, or where running a data center represented too great a cost from a power and cooling perspective. The new offices were “the trigger event to rethink what we’re doing and get out of the business” of maintaining data centers in-house, Stroot said.