Page 210 -
P. 210

10: Green IT Case Studies for Universities and a Large Company  175



                0.70
                                     1-Way: Mean  2-Way: Mean  4-Way: Mean
                   0.59
                0.60  0.52
               Fraction of 1500 Systems  0.40  0.46
                0.50




                0.30

                      0.19
                        0.17
                0.20
                       0.17
                          0.10  0.10
                0.10     0.07  0.06
                            0.04  0.05  0.03  0.04  0.06  0.02  0.03  0.05  0.01  0.02  0.03  0.01  0.02  0.02  0.01  0.01  0.02  0.01
                0.00                       0.00  0.01  0.00  0.01  0.01  0.01  0.00  0.01  0.01  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.01  0.00
                   0-5  5-10  10-15  15-20  20-25  25-30  30-35  35-40  40-45  45-50  50-55  55-60  60-65  65-70  70-75  75-80  80-85  85-90  90-95  95-100
                                       Utilization Range Bin
             Source: IBM Green Data Center
                                                                                                               ptg
             Figure 10.4 Original distribution of CPU utilization for one-way, two-way, and four-way UNIX physical
             servers
                Even if only 10 percent of your servers average less than 5 percent monthly
             utilization, there is still room for improvement. One technique used to mini-
             mize cost on the larger multiprocessor servers was to use capacity-on-demand
             technology. Capacity-on-demand technology allows for servers with extra
             capacity that can be turned on if the application requires additional
             resources. This keeps down the cost without increasing the risk of running
             out of resources to run an application.
                One of the accounts in the data center has been aggressively pursuing vir-
             tualization over the last two years. This included senior management support
             of the default direction to virtualize all workloads unless architecturally con-
             strained from doing so. This account had more than a thousand servers.
             Approximately two-thirds of the virtual images were located in the subject
             data center for their test and development workload. Their production
             servers were in another data center. The account leveraged virtualization
             technology to use the test and development site to be the backup for the pro-
             duction site.
                In the UNIX environment, a large number of 16-processor servers were
             installed on this account, leveraging capacity on-demand technology to keep
             four processors turned off for future growth. The account was originally risk
   205   206   207   208   209   210   211   212   213   214   215