Page 82 - Cloud Essentials
P. 82

Depending on the application load, more cores might improve performance.

               Some  applications  require  vast  amounts  of  memory  and  little  CPU
               processing power, and other applications might require the opposite or both.
               On  the  client  side,  cloud  computing  does  not  require  significant  CPU
               processing  power  since  most  SaaS  cloud  apps  are  provided  in  a  web
               browser.


               Storage Access Speed
                 Another challenge with cloud storage is that it is never as fast as local disks

               can be. This challenge is likely to grow worse over time since data volumes
               are  growing  faster  than  network  bandwidth.  This  is  relevant  in  normal
               operation, which includes cloud backups, and when performing imports and
               exports. For normal operation, the implication is that data storage must be

               physically close to any heavy processing that must be done with it. Moving
               either large data sets or complete virtual machines in and out of a particular
               cloud is not to be taken lightly.



               Data Replication
               When  using  cloud  computing,  applications  are  likely  to  be  spread  over
               multiple providers, whether for technical, business, or regulatory reasons. It
               then becomes a new challenge to keep the master data of these applications
               synchronized.  Some  examples  of  master  data  are  customer  account
               information, catalog data, and user identities. A number of approaches are

               possible for realizing the necessary data synchronization.


                      Replicate a master copy of the data to all applications hosted by all
                     providers. This requires careful attention to how often replication
                     occurs.
                      Synchronize all data changes in all directions when changes occur.
                     This can be difficult to implement as it may require the use of

                     additional tools.
                      Rather than replicating data, data can be looked up in real time at a
                     central source. The benefit is that the data is always up to date, but the

                     drawback is network latency.



               Data Integration
               When using cloud services, it is possible for cloud software to communicate

               with on-premises software. Integrating data in real time in a hybrid cloud
               environment once again depends on the quality and speed of your network



                                                           82
   77   78   79   80   81   82   83   84   85   86   87