Harvard Medical School Deploys InfraStruXure to Safeguard ˜On Demand" Applications

"The InfraStruXure system has far exceeded our expectations. Now that I look back, I don't know how we ever operated without it."

John Halamka, CIO (pictured)
Stephen Martino, Director of Production Operations
Harvard Medical School

At Harvard Medical School we have established ourselves as one of the first all-digital medical schools in the US. Every one of our lectures is captured electronically, streamed and archived for playback on-demand. Our microscope slides are imaged and stored so our students can focus on histology and pathology observations and not on the mechanical workings of the microscope. We also provide on-demand access to 8,000 medical journals anywhere, anytime.

Our IT staff is tasked to deliver scalable, cutting-edge technology applications 24x7x365. One area that we've focused on has been the standardization of our IT infrastructure. We have determined that a standardized approach to technology deployment helps to reduce human error, allows us to detect the root cause of defects, and creates shared learning across the organization.

Standardization is a Key Critical Success Factor

To better support our students, faculty and researchers, we constructed a new data center designed around our standardization philosophy. We performed a thorough assessment of available power, rack and cooling solutions. Most of the solutions we investigated were comprised of a series of unique point products, manufactured by multiple vendors, and not designed to work together.

These traditional systems are capable of alerting us to faults only after the damage is done. The task of identifying and correcting defects is difficult with these solutions and, because multiple vendors are involved, the equipment lacks standardization. In addition, we found that support for these traditional solutions requires a heavy dependence on outside specialists.

InfraStruXure Approach Mirrors Harvard IT Philosophy

APC's InfraStruXure data center architecture featured the standardized approach we were looking for. With the InfraStruXure system, we immediately receive alerts over the network if a threshold is exceeded, and the system identifies, on its own, exactly where the fault is occurring. The system is architected with internal redundancy and stays up and running despite the fault. We can self-remedy the situation by simply removing and replacing hot-swappable components. InfraStruXure was the only data center power system we encountered whose fundamental architecture was based on pre-engineered and pre-tested building blocks designed to work in a truly integrated fashion.

High Density Enters into the Equation

Support for high density computing was also a key consideration in selecting the InfraStuXure.  To minimize real-estate costs we needed to grow our IT equipment footprint vertically rather than horizontally. We needed to pack more highly dense blade servers into each rack. We had always faced power density issues in our old data center where we averaged 1.5 kW of power per rack. As we began to deploy blade servers and storage we were approaching an average requirement of 3-5 kW per rack.

The traditional rack and power distribution schemes we looked at could only accommodate, on average, 1-2 kW per rack. The design of the InfraStruXure racks allowed for zero power distribution and enables us to install up to 42 blades in one rack. We can now pack more than 5kW of blade servers per rack.

Benefits Include Balanced Weight Distribution and Floor Space Savings

The overhead power distribution design enables us to more effectively distribute cooling because the space below our raised floor is no longer stuffed with cables that hinder proper air distribution.

If we need to, we can move the system to a new location without sacrificing any of our initial investment.  The weight of the system is not an issue. A flexible floor plan allows us to spread the weight loads across the floor instead of concentrating the weight all in one dense spot. The InfraStruXure system has far exceeded our expectations.  Now that I look back, I don't know how we ever operated without it.