Page 205 -
P. 205

The Greening of IT
           170                  How Companies Can Make a Difference for the Environment



           requires physical resources. That is, until they run out. In the early days of
           distributed computing, many IT organizations took the throw-another-server-
           at-it approach because servers were inexpensive compared to the proper capac-
           ity planning needed to optimize hardware equipment utilization.
             Back in the 1980s, the author of this case study, Chris Molloy, was person-
           ally involved in this approach. It wasn’t until the financial approval for the
                st
           2,501 server was requested that the finance team asked if there were room
           for the workload on the first 2,500 servers. The answer was not available as
           capacity management tools were not installed at that time. In agreement to
           provide funding for future server requests, we agreed to install capacity man-
           agement tools and to demonstrate that utilization of existing servers justified
           the investment in new servers for new workload. Even so, the multitier dis-
           tributed applications ran at low utilization because capacity planners were
           required to size servers for peak utilization versus average utilization. In
           those days, space, power, and cooling were not an issue because improve-
           ments in miniaturization of IT resources exceeded the physical growth
           requirements.
             Much has changed since those days. Virtualization techniques have
                                                                                                            ptg
           allowed companies to share reserve capacity between multiple workloads,
           increasing server utilization. There has been a rapid increase in the demand
           for IT resources, with current studies showing a 6X growth in servers and
           69X growth in storage resources in this decade. The cost dynamics have
           changed, with electricity costs in the data center increasing in double-digit
           percentages for the past two years. Equipment manufacturers have responded
           to these challenges by making larger, more-efficient hardware.
           Unfortunately, the physical characteristics of power, space, and cooling have
           become an issue because previous data centers cannot contain the new physi-
           cal requirements of these larger servers or cannot contain the volume of
           servers needing to be installed for new IT requirements.
             Such was the case with the data center that is the subject of this study. To
           install a new piece of IT equipment, you need to have the space to put it in,
           the power to run it, and the cooling to dissipate the heat generated from the
           power consumed. The installation of a piece of equipment requires power,
           space, and cooling. Additionally, the applications had a high-availability
           requirement. Each watt of utility power was backed up by UPS, batteries
           that provide power for about 15 minutes. This time allows for longer-term
           generators to start and stabilize until utility power resumes. Any of these
   200   201   202   203   204   205   206   207   208   209   210