Inside-Out: Where Real Optimization Begins

In 2005, one of Dell's data centers in the Austin area was comprised of 6,000 sf of raised floor with around 140 racks of IT equipment. Without space or power for more racks, the infrastructure team conducted an IT assessment and determined the actual workload performed. A virtualization and consolidation effort was begun and by 2009, the same workload was being performed by 40 racks in 1,600 sf. The same workload can be performed today (in 2013) by equipment that will fit into two 54U racks. In 2012, Dell migrated completely out of that data center and sold it, consolidating several sites into one data center.

While the drive to make infrastructure more efficient over the last 5 years has changed the conversation with respect to data center design and made designers and owners more interested in "green" initiatives, there is a theoretical limit to what can be achieved solely through infrastructure savings. A high density environment like the one described above, may consume approximately 25kW, or about $1,800 per month in energy costs just for the equipment. In order to cool the environment to 70 deg F, a normal cooling system might cost another $1,000 per month to operate. Over the three years of a typical IT platform lifecycle, the total energy outlay would be just over $100,000. Compared to the IT equipment cost - which can run anywhere from $10,000 to $20,000 per rack unit - the energy cost is about 10% to 20% of the total lifecycle cost of $500k to $1 million. Even if we eliminated the energy used to cool the equipment, the savings would be in a very pedestrian 10% to 20% range and further reductions would not be possible.

This is where enterprise optimization becomes the driver of innovation that generates game changing results. By re-architecting the enterprise, Dell was able to completely eliminate an entire data center and its multi-million dollar annual carrying costs. While not all workloads will be able to immediately be migrated to a virtual environment, any optimization effort must provide visibility into the ecosystems and platforms comprising the business critical applications and their hardware dependencies. There are specific strategies and tactics that are now well defined that can begin to deal with these IT workloads that comprise the inside of the data center, and then allow the outside - the core and shell, power distribution, and cooling systems - to be right-sized and optimized for both reliability and efficiency to support the enterprise at the lowest possible cost.