Page 1044 - Cloud computing: From paradigm to operation
P. 1044

7                                                     Security


            −       the storage location including local and/or remote;
            −       the retention periods for backup data;

            −       the procedures to test the backup data.
            Before  choosing  CSP,  CSC  should  confirm  whether  that  CSP  could  meet  the  security  clause  of  the  SLA
            including the capability of backup. If CSP does not provide a backup capability, CSC should fully consider a
            backup  strategy  and  implementation.  Otherwise,  if  CSP  provides  a  backup  capability,  then  CSC  should
            cooperate with CSP to carry out backup operations.

            CSP should share the essential details of the backup mechanism with CSCs. When dealing with backup, CSPs
            should address the specifications to meet each of the following CSC's requirements:
            1)      Backup  strategy:  Since  each  CSC  has  individual  needs  in  backup,  the  related  factors  should  be
                    primarily considered, which include:
                    −   Reasonable recovery point objective (RPO) and recovery time objectives (RTO). RPO indicates
                        the time span between two consecutive backup activities, while RTO reflects how long it takes
                        to roll-back to a backup.
                    −   Reasonable retention policy: The policy should specify the copy number of a backup.
                    −   Reasonable combination of file-level backup and virtual machine level backup: The combination
                        should satisfy an optimal investment cost, which is based upon RPO and RTO.
                    −   Reasonable combination of on-site backup and off-site backup: The on-site backup is stored in
                        the local site, which could meet the need of fast disaster recovery. The off-site backup is stored
                        in a remote site, which is needed to cope with a major disaster. The combination depends on
                        the requirement of the security clause of the SLA, and the investment cost.

                    −  Regular test procedures of recovery: The recovery test is the ultimate method to verify the
                        validity of a backup.
            2)      Task arrangement: Once the backup strategy is determined, CSPs should make an appropriate task
                    arrangement  of  backup  operations.  To  reduce  the  impact  on  the  performance  of  the  cloud
                    computing  infrastructure,  the  backup  task  arrangement  should  depend  on  CSC's  backup
                    requirements, the network traffic pattern and the backup capability of CSP.
            3)      Procedures to check the validation of backup: A complete and correct data copy means a successful
                    backup operation. Generally, the procedures should contain the following two main steps:
                    −  Using a one-way hash function to verify that the backup is consistent with the original data. If
                        the backup is the same as the original, then go to the next step. Moreover, a digital signature
                        method could be used to verify the backup operator, which can introduce some benefit to the
                        management of backup operation.
                    −  Taking  a  recovery  test  for  the  backup.  As  the  continuous  change  in  the  cloud  computing
                        environment, regular recovery test is critical.
            4)      Prudence about the snapshots of the virtual machine: In a cloud computing scenario, the snapshot
                    method provides a quick and easy means of rollback, which could act as a backup method to a
                    certain extent. However, the snapshot method should not be used frequently due to the following
                    reasons:
                    −   Snapshots allow the same data to multiply and to be written in different snapshot files, which
                        could easily bring about serious performance degradation and rapid storage occupancy in the
                        cloud computing systems.
                    −   In order to reduce storage occupancy, a chain of an original virtual machine snapshots is often
                        configured to merely contain the difference from the first snapshot. Once the first snapshot is
                        destroyed, the successive snapshots would end up being invalid. The security risk is magnified
                        as the rate of successive snapshots increases.





            1036
   1039   1040   1041   1042   1043   1044   1045   1046   1047   1048   1049