1st TCom Meeting: 14th March 2012 Discussions and Notes

From IMarine Wiki

Jump to: navigation, search


Meeting Agenda

Meeting Participants

WP7 - Bringing Maven to gCube

G. Giammatteo (ENG)

Slides: pdf file

New services should be conceived for Maven. There is no added value in retaining the "old" and "proprietary" approach.

There is no support for portlets.

However, we should take into account that there is a learning curve for developers if they are not familiar with Maven.

T9.1 - Data Access and Storage

Fabio Simeoni (FAO)

Slides: pdf file

The infrastructure offers a number of "domain specific" data access and discovery services, e.g. time series, geospatial

  • these represents sort of "islands" of data, the "tree manager" might be a way to aggregate these islands
  • (IRD) the best for us is to enlarge the 'formats' our data are exposed

T9.2 - New Data transfer

Andrea Manzi (CERN)

Slides: pptx file

T9.2 - gRS2 extensions

Gerasimos Farantatos (NKUA)

Slides: pptx file

Credentials information can be stored in the information system by relying on the new facilities for storing encrypted information.

The library for URL resolution might be useful in many cases beyond the data transfer including WP11 activities.

Geospatial Data management

Fabrice Brito (Terradue)

Slides: ppt file

OGC WPS is implemented by 52North, the proposed approach relies on.

  • the processes (P_i) should be implemented by extending an interface designed by 52North;

WPS-hadoop is a piece of software implemented by Terradue in the context of iMarine.

Use Case #2: the step dealing with layers publishing might count on CNR facilities for storing layers in a geoserver cluster.

In theory it might be possible to deploy P_i on THREDDS service instance.

The proposal is based on Cloudera Hadoop distribution.

The interaction with myOcean is based on Fabrice "credentials".

The feeding of the "catalogue" can be based on subscriptions, i.e. when the catalogue component is configured to run in a given domain it can be configured to automatically get the data.

Action: A working group should be built in order to design the integration and deployment of the proposed approach in the context of the iMarine Infrastructure. Discussions include:

  • how to inform the WPS-hadoop about the Hadoop clusters available in the infrastructure;
  • how and when to deploy the software pieces;

Terradue is investigating the Oozie workflow tool for Hadoop.

T8.3 - PE2ng evolution

George Kakaletris (NKUA)

Slides: []

T8.1/T8.4 - The evolution of Resource Management

Fabio Simeoni (FAO)

Slides: pdf file

There might be the need to share something across diverse facets of the same resource.

Facets should be somehow agreed, there should be a coordination phase.

Checking of facets can be forced during registration/creation phase.

Further discussion is needed for: "technological targets", resources other than software;

T8.2 - Security a.k.a. SOA3

Ciro Formisano (ENG)

Slides: pdf file

There are different set of policies that requires diverse services to take care of them.

SLA policies are expected to be managed by a "monitoring service" rather than an "authorization service".

The policies to be managed should be identified, this is a pre-requisite for service design.

Policies applies to "resources", i.e. whatever is modeled as a resource in the Information System.

Manuele (CNR): the requirement on "more flexibility on policies" should include the possibility to define the actions on which policies can be defined;

  • ENG: this is the current behaviour, i.e. the authorization framework just replies if a given action on a target resource is allowed or not;

For user management and diverse databases:

  • Liferay can be configured to store data in an LDAP DB;

Steps to be discussed:

  • how this approach fits with the "zero-management" approach introduced during previous talk;
  • identify custom attributes;
  • SLA vs AuthZ policies;
  • it should be discussed whether the point raised by Manuele is there or not;
  • the maintenance of the Liferay DB and sync with LDAP;

T10.1/T10.2 - Data Retrieval and Manipulation

Gerasimos Farantatos (NKUA)

Slides: pptx on retrieval pptx on manipulation

There has been a refactoring of the services, namely the Index part.

Selective update of indices is still in the plan by relying on subscription/modification, however this is not a top priority.

T10.3 - Data Mining

Gianpaolo Coro (CNR)

Slides: []

An important issue is to decide the level of exploitation of/ integration with PE2ng to enlarge the processing part.

No algorithm for clustering is integrated yet. NKUA is willing to propose approaches for this.

NKUA is also willing to contribute to data visualization.

T10.4 - Semantic Data Analysis

Yannis Tzitzikas (FORTH)

Slide: pdf file

X-Search should take into account the different contexts in which it can work, eg it should be configured to support the discovery on a number of data sources. As a consequence, the capabilities of X-Search might change.

Moreover, the whole implementation should be analysed from an infrastructure POV.

Re provenance, we should understand what the community actually needs. We can count on CIDOC-CRM (a few classes of this very rich model);

Action: to discuss on the real integration plan (including integration from the software pov and from the infrastructure pov).

Personal tools