Page 963 - Cloud computing: From paradigm to operation
P. 963

Monitoring                                                  6


            See Table 7-6.


                                 Table 7-6 – Parameters collected to monitor a service chain

                        Metric name                             Description                       Unit

             NFS                               Network function status                          normal/fail
             NF start                          Time of network function start                       s
             NF stop                           Time of network function stop                        s
                                               Time that is needed to move a NF from two
             NF migration                                                                           s
                                               predefined resources
             SC start                          Time of a service chain established                  s
             SC stop                           Time of a service chain destroy                      s

            7.1.6   Energy consumption
            Voltage or power use out of the permissible range can damage electrical components or cause system failure.
            If the fan stops working, the server overheats, is damaged and goes out of service. So, it is important to
            monitor the voltage or wattage, fan and temperature to ensure that they are within safe operating limits.

            In  order  to  handle  massive  amounts of  data  generated  by  consumers  and  businesses,  cloud  computing
            typically needs a lot of power. A sharp increase in energy consumption can indicate server load unbalance
            that leads to performance degradation. Real-time monitoring is helpful for avoiding these consequences.
            Different states of the physical or VMs require different power levels. These states can normally be divided
            into six types: named as shutdown, work, idle, dormant, sleep and standby.

            See Table 7-7.


                         Table 7-7 – Parameters collected to monitor energy consumption in each state

                        Metric name                            Description                       Unit
             power of CPU                      Current power of CPU consumption                    w
             power of GPU                      Current power of GPU consumption                    w
             power of Memory                   Current power of Memory consumption                 w

             power of storage                  Current power of storage consumption                w
             power of network                  Current power of disk consumption                   w
             power of power Systems            Current power of power system consumption           w
             temperature of CPU                Current temperature of CPU                         o C

            7.1.7   Environment
            The server room environment requires strict control processes for temperature, humidity and power supply.
            A  fully  automated  monitoring  system  can  help  prevent  overheating  of  servers  and  condensation  on
            equipment. Keeping temperatures within range and carefully monitoring humidity to prevent corrosion or
            static electricity reduce energy consumption and keep servers running smoothly. Voltage use out of the
            permissible range can damage electrical components or cause system failure. If the fan stops working, the
            server overheats, is damaged and goes out of service. So, it is important to monitor the voltage or wattage,
            fan and temperature to ensure that they are within safe operating limits.












                                                                                                         955
   958   959   960   961   962   963   964   965   966   967   968