Transformation of a business often requires a transformation of its technology infrastructure. For Barry Novick, global data center manager at BlackRock, an acute need for such a shift came when, through an acquisition, BlackRock became the world’s largest fund manager.

Once the US$13.5bn acquisition of Barclays Global Investors (Barclays Plc’s investment management business) was closed in December 2009, the asset portfolio under BlackRock’s management grew to $2.7 trillion. According to Bloomberg, that was more than the value of assets managed by the US Federal Reserve.
 

By the time BlackRock made the Barclays deal, a new technology strategy was in the planning stages, Novick says. The inheritance of about a dozen data centers through the acquisition put the infrastructure overhaul on fast track.

The company had another 12 prior to the deal, many received through prior acquisitions These facilities were constructed at different times under different technology models, and there was clearly a lack of uniformity. “So we spent 2011 consolidating, shutting down and migrating,” Novick says. As a result, the

company went back to about 12 data centers, four of which were new.

One of the new facilities is on a Sabey property in East Wenatchee, Washington. Now BlackRock’s primary data center, the 3.2MW site is a crown jewel of its technology strategy. From using indirect evaporative cooling instead of traditional chiller-based air conditioning to overhead cooling instead of raised floor, the data center is a symbol of the new era for BlackRock’s technology infrastructure.
 

It was built faster and cheaper than ever. One major way BlackRock cut costs was by rethinking its application uptime requirements. 

Smarter Risk Management

Fundamentally, BlackRock is in the business of risk management. Its risk analytics applications consume a majority of its data center resources, and with the new technology strategy the company also took a smarter approach to managing its data center downtime. One key element of the strategy is a case-by-case approach to infrastructure resiliency requirements. The company consolidated applications and re-examined each application’s reliability needs. “Not everything requires the same resiliency,” Novick says.

BlackRock’s most critical applications are supported by highly redundant infrastructure, and services that are more disposable are supported by lower-tier topologies.

Instead of designing infrastructure to have zero risk, BlackRock’s aim is “to have the right amount of risk at the right price and understand that relationship”, Novick says.

Right-sizing redundancy levels based on criticality of specific applications can cut costs significantly. In a 2009 white paper, Eaton’s data center applications manager 

Chris Loeffler wrote: “Generally speaking, the higher the tier, the more power resources you require. So it is imperative that organizations weigh the additional costs of deploying a higher-tier design against their business objectives and service level agreements.” 

Away with raised floor

Another major change in BlackRock’s strategy is a move away from using raised floor across its entire data center footprint. Instead, where possible, the company uses a combination of overhead cooling and contained PODs.

“Raised flooring has a fixed cost and if our cooling architecture does not require it, we can save that money,” Novick says. In addition to cost savings, there are engineering reasons for not submitting to the raised floor orthodoxy.

While the majority of data centers continue to be built with raised floor, there are a growing number of designs that break away with the tradition. In a 2011 white paper,

Neil Rasmussen, senior VP of innovation at Schneider Electric, said data room

requirements that led to invention of raised floor have changed, so it is no longer the best solution for every site.

First of all, power densities in today’s data centers are much higher than they were when raised floor became the de facto standard. A  lot of modern IT equipment operates at 25kW or more per rack, which requires four times the airflow that traditional raised floor was designed for, Rasmussen says.

Another key difference is the dynamism of today’s data center environment. Data centers in the past supported equipment that was not replaced or moved frequently, thus requiring no changes to cooling. Today, cooling requirements change constantly. A modern data center will see its IT infrastructure refreshed multiple times over its lifespan.

BlackRock isn’t dogmatic in its use of overhead cooling. “A lot has to do with the available architecture of the sites,” Novick says. Overhead cooling requires high ceilings, which aren’t available everywhere, so traditional raised floors are still used wherever they are necessary. 

Variety is the only constant

No one-size-fits-all is a running theme throughout BlackRock’s current strategy.

Each infrastructure decision is guided by case-specific requirements. Take, for

example, the company’s assortment of data center types – “from owned and operated to retail colocation, and everything in between,” as Novick puts it.

From the massive flagship facility in Washington State to a colocation cage in

Europe, the new strategy is working.

Novick says it will continue into the future because it resulted in an infrastructure that is “much more efficient and much more resilient.”