Metacat

DataONE Member Node Support

===========================

DataONE_ is a federation of data repositories that aims to improve 

interoperability among data repository software systems and advance the

preservation of scientific data for future use.

Metacat deployments can be configured to participate in DataONE_. This 

chapter describes the DataONE_ data federation,  its architecture, and the 

way in which Metacat can be used to participate as a node in the DataONE system.

.. _DataONE: http://dataone.org/

What is DataONE?

----------------

The DataONE_ project is a collaboration among scientists, technologists, librarians,

and social scientists to build a robust, interoperable, and sustainable system for

preserving and accessing Earth observational data at national and global scales.  

Supported by the U.S. National Science Foundation, DataONE partners focus on

technological, financial, and organizational sustainability approaches to 

building a distributed network of data repositories that are fully interoperable,

even when those repositories use divergent underlying software and support different

data and metadata content standards. DataONE defines a common web-service service 

programming interface that allows the main software components of the DataONE system

to seamlessly communicate. The components of the DataONE system include:

* DataONE Service Interface

* Member Nodes

* Coordinating Nodes

* Investigator Toolkit

Metacat implements the services needed to operate as a DataONE Member Node, 

as described below.  The service interface then allows many different scientific 

software tools for data management, analysis, visualization and other parts of 

the scientific lifecycle to directly communicate with Metacat without being

further specialized beyond the support needed for DataONE.  This streamlines the

process of writing scientific software both for servers and client tools.

The DataONE Service Interface

-----------------------------

DataONE acheives interoperability by defining a lightweight but powerful set of 

REST_ web services that can be implemented by various data management software 

systems to allow those systems to effectively communicate with one another, 

exchange data, metadata, and other scientific objects.  This `DataONE Service Interface`_

is an open standard that defines the communication protocols and technical 

expectations for software components that wish to participate in the DataONE

federation. This service interface is divided into `four distinct tiers`_, with the 

intention that any given software system may implement only those tiers that are 

relevant to their repository; for example, a data aggregator might only implement

the Tier 1 interfaces that provide anonymous access to public data sets, while

a complete data management system like Metacat can implement all four tiers:

1. **Tier 1:** Read-only, anonymous data access

2. **Tier 2:** Read-only, with authentication and access control

3. **Tier 3:** Full Write access

4. **Tier 4:** Replication target services

.. _REST: http://en.wikipedia.org/wiki/Representational_state_transfer

.. _DataONE Service Interface: http://releases.dataone.org/online/d1-architecture-1.0.0

.. _four distinct tiers: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/index.html

Member Nodes

------------

In DataONE, Member Nodes represent the core of the network, in that they represent

particular scientific communities, manage and preserve their data and metadata, and

provide tools to their community for contributing, managing, and accessing data.

DataONE provides a standard way for these individual repositories to interact, and helps

to coordinate among the Member Nodes in the federation.  This allows Member Nodes

to provide services to each other, such as replication of data for backup and failover.

To be a Member Node, a repository must implement the Member Node service interface, 

and then register with DataONE.  Metacat provides this implementation automatically,

and provides an easy configuration option to register a Metacat instance as a 

DataONE Member Node (see configuration section below). If you are deploying a Metacat

instance, it is relatively simple to become a Member Node, but keep in mind that 

DataONE is aiming for longevity and preservation, and so is selecting for nodes

that have long-term data preservation as part of their mission. 

Coordinating Nodes

------------------

The DataONE Coordinating Nodes provide a set of services to Member Nodes that

allow Member Nodes to easily interact with one another and to provide a unified

view of the whole DataONE Federation.  The main services provided by Coordinating

Nodes are:

* Global search index for all metadata and web portal for data discovery

* Resolution service to map unique identifiers to the Member Nodes that hold data

* Authentication against a shared set of accounts based on CILogon_ and InCommon_

* Replication management services to reliably replicate data according to 

  policies set by the Member Nodes

* Fixity checking to ensure that preserved objects remain valid

* Member Node registration and management

* Aggregated logging for data access across the whole federation

Three geographically distributed Coordinating Nodes replicate these coordinating 

services at UC Santa Barbara, the University of New Mexico, and the Oak Ridge Campus.

Coordinating Nodes are set up in a fully redundant manner, such that any of the coordinating

nodes can be offline and the others will continue to provide availability of the services

without interruption.  The DataONE services expose their services at::

  https://cn.dataone.org/cn

And the DataONE search portal is available at:

  https://cn.dataone.org/

.. _CILogon: http://www.cilogon.org

.. _InCommon: http://incommon.org

Investigator Toolkit

--------------------

In order to provide scientists with convenient access to the data and metadata in

DataONE, the third component represents a library of software tools that have been 

adapted to work with DataONE via the service interface and can be used to

discover, manage, analyze, and visualize data in DataONE.  For example, DataONE

plans to release metadata editors (e.g., Morpho), data search tools (e.g., Mercury), 

data access tools (e.g., ONEDrive), and data analysis tools (e.g., R) that all 

know how to interact with DataONE Member Nodes and Coordinating Nodes.  Consequently,

scientists will be able to access data from any DataONE Member Node, such as a Metacat

node, directly from within the R environment.  In addition, software tools that 

are written to work with one Member Node should also work with others, thereby

greatly increasing the efficiency of creating an entire toolkit of software that

is useful to investigators.  

Because DataONE services are REST web services, software written in any

programming language can be adapted to interact with DataONE.

In addition, to ease the process of adapting tools to work with DataONE, libraries

are provided for common programming languages such as Java (d1-libclient-java) 

and Python (d1_libclient-python) are provided that allow simple function calls 

to be used to access any DataONE service.

Configuring Metacat as a Member Node

------------------------------------

Configuring Metacat as a DataONE Member Node is accomplished with the standard

Metacat Administrative configuration utility. To access the utility, visit the 

following URL::

  http://<yourhost.org>/<context>/admin

where ``<yourhost.org>`` represents the hostname of your webserver running metacat,

and ``<context>`` is the name of the web context in which Metacat was installed.

Once at the administrative utility, click on the DataONE configuration link, which

should show the following screen:

.. figure:: images/screenshots/screen-dataone-config.png

   :align: center

   The configuration screen for configuring Metacat as a DataONE node.

To configure Metacat as a node in the DataONE network, configure the properties shown

in the figure above.  The Node Name should be a short name for the node that can

be used in user interface displays that list the node.  For example, one node in

DataONE is the 'Knowledge Network for Biocomplexity'.  Also provide a brief sentence

or two describing the node, including its intended scope and purpose.  

The Node Identifier field is a unique identifer assigned by DataONE to identify

this node even when the node changes physical locations over time.  When Metacat

registers with the DataONE Coordinating Nodes (when you click 'Register' at the

bottom of this form), the Node Identifier is automatically set.  **It is critical that

you not change the Node Identifier**, as that will break the connection with the

DataONE network.  The ability to edit this field is only provided for the rare case

in which a new Metacat instance is being established to act as the provider for an 

existing DataONE Member Node, in which case the field can be edited to set it to

the value of a valid, existing Node Identifier.

The Node Subject and Node Certificate Path are linked fields that are critical for

proper operation of the node.  To act as a Member Node in DataONE, you must obtain

an X.509 certificate that can be used to identify this node and allow it to securely

communicate using SSL with other nodes and client applications.  This certificate can 

either be obtained from the DataONE Certificate Authority, or from a commercial 

provider of certificates. Once you have the certificate in hand, use a tool such 

as ``openssl`` to determine the exact subject distinguished name in the 

certificate, and use that to set the Node Subject field.  Set the Node 

Certificate Path to the location on the system in which you stored the 

certificate file.

The ``Synchronize`` checkbox allows the administrator to decide whether to turn on

synchronization with the DataONE network.  When this box is unchecked, the DataONE

Coordinating Nodes will not attempt to synchronize at all, but when checked, then

DataONE will periodically contact the node to synchrnize all metadata content.

To be part of the DataONE network, this box must be checked as that allows 

DataONE to receive a copy of the metadata associated with each object in the Metacat

system.  The switch is provided for those rare cases when a node needs to be disconnected

from DataONE for maintenance or service outages.  When the box is checked, DataONE

contacts the node using the schedule provided in the ``Synchronization Schedule``

fields.  The example in the dialog above has synchronization occurring once every third

minutes at the 10 second mark of those minutes.  The syntax for these schedules

follows the Quartz Crontab Entry syntax, which provides for many flexible schedule 

configurations.  If the administrator desires a less frequent schedule, such as daily, 

that can be configured by changing the ``*`` in the ``Hours`` field to be a concrete 

hour (such as ``11``) and the ``Minutes`` field to a concrete value like``15``, 

which would change the schedule to synchronize at 11:15 am daily.  

Once these parameters have been properly set, us the ``Register`` button to

request to register with the DataONE Coordinating Node.  This will generate a

registration document describing this Metacat instance and send it to the 

Coordinating Node registration service, which will return a unique Node Identifier

which will be recorded by Metacat.  At that point, all that remains is to wait for

the DataONE administrators to approve the node registration.  Details of the approval

process can be found on the `DataONE web site`_.

.. _DataONE web site: http://www.dataone.org

Access Control Policies

-----------------------

Metacat has supported fine grained access control for objects in the system since

its inception.  DataONE has devised a simple but effective access control system

that is compatible with the prior system in Metacat.  For each object in the DataONE

system (including data objects, scientific metadata objects, and resource maps), 

a SystemMetadata_ document describes the critical metadata needed to manage that

object in the system.  This metadata includes a ``RightsHolder`` field and an

``AuthoritativeMemberNode`` field that are used to list the people and node that

have ultimate control over the disposition of the object.  In addition, a separate

AccessPolicy_ can be included in the ``SystemMetadata`` for the object.  This ``AccessPolicy``

consists of a set of rules that grant additional permissions to other people, 

groups, and systems in DataONE.  For example, for one data file, two users 

(Alice and Bob) may be able make changes to the object, and the general public may

be allowed to read the object.  In the absence of explicit rules extending these permissions,

Metacat enforces the rule that only the ``RightsHolder`` and ``AuthoritativeMemberNode`` have

rights to the object, and that the Coordinating Node can manage ``SystemMetadata``

for the object.  An example AccessPolicy that might be submitted with a dataset

(giving Alice and Bob permission to read and write the object) follows:

::

  ...

  <accessPolicy>

      <allow>

        <subject>/C=US/O=SomeIdP/CN=Alice</subject>

        <subject>/C=US/O=SomeIdP/CN=Bob</subject>

        <permission>read</permission>

        <permission>write</permission>

      </allow>

  </accessPolicy>

  ...

These AccessPolicies can be embedded inside of the ``SystemMetadata`` that accompany

submission of an object through the `MNStorage.create`_ and `MNStorage.update`_ services, 

or can be set using the `CNAuthorization.setAccessPolicy`_ service.

.. _SystemMetadata: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/Types.html#Types.AccessPolicy

.. _AccessPolicy: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/Types.html#Types.AccessPolicy

.. _MNStorage.create: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/MN_APIs.html#MNStorage.create

.. _MNStorage.update: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/MN_APIs.html#MNStorage.update

.. _CNAuthorization.setAccessPolicy: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/CN_APIs.html#CNAuthorization.setAccessPolicy

Configuration as a replication target

-------------------------------------

DataONE is designed to enable a robust preservation environment through replication

of digital objects at multiple Member Nodes.  Any Member Node in DataONE that implements

the Tier 4 Service interface can offer to act as a target for object replication.  

Currently, Metacat configuration supports turning this replication function on or off.

When the 'Act as a replication target' checkbox is checked, then Metacat will notify

the Coordinating Nodes in DataONE that it is available to house replicas of objects

from other nodes.  Shortly thereafter, the Coordinating Nodes may notify Metacat to

replicate objects from throughout the system, which it will start to do.  There objects

will begin to be listed in the Metacat catalog.

.. Note:: 

  Future versions of Metacat will allow finer specification of the Node

  Replication Policy, which determines the set of objects

  that it is willing to replicate, using constraints on object size, total objects, 

  source nodes, and object format types.

Object Replication Policies

---------------------------

In addition to access control, each object also can have a ``ReplicationPolicy``

associated with it that determines whether DataONE should attempt to replicate the

object for failover and backup purposes to other Member Nodes in the federation. 

Both the ``RightsHolder`` and ``AuthoritativeMemberNode`` for an object can set the

``ReplicationPolicy``, which consists of fields that describe how many replicas 

should be maintained, and any nodes that are preferred for housing those replicas, or

that should be blocked from housing replicas.  

These ReplicationPolicies can be embedded inside of the ``SystemMetadata`` that accompany

submission of an object through the `MNStorage.create`_ and `MNStorage.update`_ services, 

or can be set using the `CNReplication.setReplicationPolicy`_ service.

.. _CNReplication.setReplicationPolicy: http://releases.dataone.org/online/d1-architecture-1.0.0/apis/CN_APIs.html#CNReplication.setReplicationPolicy