120 likes | 288 Views
EUDAT Data Policy Manager . Mark van de Sanden (SURFsara) Maria Francesca Iozzi (SIGMA/University of Oslo) Claudio Cacciari (CINECA) RDA 3 rd Plenary meeting Practical Policy Working Group. Date : 27 March 2014. Objectives.
E N D
EUDAT Data Policy Manager Mark van de Sanden (SURFsara) Maria Francesca Iozzi (SIGMA/University of Oslo) Claudio Cacciari (CINECA) RDA 3rd Plenary meeting Practical Policy Working Group Date: 27 March 2014
Objectives • allow a Community Manager (CM) to specify data management policies • allow a Community Manager (CM) to manage those policies via a web-portal
Replication 1839/abc EUDAT 29db...279b4a 1.10.13 00:00 CM wants to specify the type of replication, the replica sites and how often data shall be replicated Replication from A to B and from B to C Replication from A to B and from A to C 456/abc 789/abc 29db...279b4a 29db...279b4a 1.10.13 02:00 1.10.13 02:00
Integrity 1839/abc EUDAT 29db...279b4a 1.10.13 00:00 CM wants to specify the periodicity of integrity checking 456/abc 789/abc 29db...279b4a 29db...279b4a 1.10.13 02:00 5.10.13 05:00 Checksum recalculation on the physical file
Data Policy Manager Portal By using the portal the CM can: • Select policies (replication, integrity checking, retention, etc…) • Hand-pick the sites involved in the policy • Visualize the history of the issued policies on a searchable table The created policies are in an abstract way, i.e. independent from any technology or technology format (e.g. iRODS rules).
Policy hierarchy Policy schema: the xml schema whichdefines the elementsof a policy document Policy schema Policy template 1 Policy template 2 Policy template 3 Policy template n … Policy Concrete Instance 1A Policy Concrete Instance 2A Policy Concrete Instance 3A Policy Concrete Instance nA Policy Concrete Instance 1B Policy Concrete Instance 2B … Policy Concrete Instance 2C
Abstract Policy template Policy template: the policy documentwhichdefines a policy process, butwithoutspecificparameters. Therefore the tasks are defined, butwithout input/output parameters. Forexample: <dataset> <collectionid="0"> <persistentIdentifiertype="PID“></persistentIdentifier> </collection> </dataset> <actions> <actionname="replicationonchange"> <type>replicate</type> <trigger> <action>modifyobject</action> </trigger> <targets> <target id="0"> <location xsi:type="irodsns:coordinates"> <irodsns:site type="EUDAT"></irodsns:site> <irodsns:path></irodsns:path> <irodsns:resource></irodsns:resource> </location> </target> </targets> </action> </actions> Define data sets Policy type Define action Target descriptons
Concrete Policy description Policy instance: the policy documentwhichdefines a policy process, butwithspecificparameters. <dataset> <collectionid="0"> <persistentIdentifiertype="PID“> 11100/6c8ac19e-c982-11e2-b3cb-e41f13eb41b2 </persistentIdentifier> </collection> </dataset> <actions> <actionname="replicationonchange"> <type>replicate</type> <trigger> <action>modifyobject</action> </trigger> <targets> <target id="0"> <location xsi:type="irodsns:coordinates"> <irodsns:site type="EUDAT">CINECA</irodsns:site> <irodsns:path>/path/to/destination</irodsns:path> <irodsns:resource>defaultResc</irodsns:resource> </location> </target> </targets> </action> </actions>
Data Policy Manager Portal Simple Replication, Replication from A to B and from A to C, Replication from A to B and from B to C, .. . How often should data be replicated Different protocols can be adopted (not only iRODS). CM can hand-pick the target sites
EUDAT is working on We are implementing a web-based portal to manage policies: • lightweight and easy to use tool • independent from any protocols • integrated with other EUDAT service (first and foremost with the B2SAFE service) • Authentication and authorization mechanism will be implemented to guarantee security. • Policies must be flexible to be tailored to Communities requirements. • More communities involved, more complex rules/policy-templates implemented.
Requirements EUDAT: Data Policy Manager Currently basic HTTP get EUDAT: iRODS! − We need a structured way to define abstract policies: policies templates, policy terminology − Define policies in a flexible way: not everything is a policy, some are actions or functions of a policy − We also need • Tools to manage policies, • A mechanism to distribute policies, • Technologies adapted to work on policies