/**
Copyright (C) SYSTAP, LLC DBA Blazegraph 2006-2016. All rights reserved.
Contact:
SYSTAP, LLC DBA Blazegraph
2501 Calvert ST NW #106
Washington, DC 20008
licenses@blazegraph.com
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/*
* Created on May 19, 2007
*/
package com.bigdata.rdf.store;
import java.util.Properties;
import com.bigdata.journal.IIndexManager;
import com.bigdata.relation.locator.DefaultResourceLocator;
import com.bigdata.service.IBigdataFederation;
/**
* Implementation of an {@link ITripleStore} as a client of an
* {@link IBigdataFederation}.
*
* <h2>Deployment choices</h2>
*
* The {@link JiniFederation} provides a scale-out deployment. The indices will
* be dynamically key-range partitioned and will be automatically re-distributed
* over the available nodes in the cluster.
*
* <h2>Architecture</h2>
*
* The client uses unisolated writes against the lexicon (terms and ids indices)
* and the statement indices. The index writes are automatically broken down
* into one split per index partition. While each unisolated write on an index
* partition is ACID, the indices are fully consistent iff the total operation
* is successful. For the lexicon, this means that the write on the terms and
* the ids index must both succeed. For the statement indices, this means that
* the write on each access path must succeed. If a client fails while adding
* terms, then it is possible for the ids index to be incomplete with respect to
* the terms index (i.e., terms are mapped into the lexicon and term identifiers
* are assigned but the reverse lookup by term identifier will not discover the
* term). Likewise, if a client fails while adding statements, then it is
* possible for some of the access paths to be incomplete with respect to the
* other access paths (i.e., some statements are not present in some access
* paths).
* <p>
* Two additional mechanisms are used in order to guarantee reads from only
* fully consistent data. First, clients providing query answering should read
* from a database state that is known to be consistent (by using a read-only
* transaction whose start time is the globally agreed upon commit time for that
* database state). Second, if a client operation fails then it must be retried.
* Such fail-safe retry semantics are available when data load operations are
* executed as part of a map-reduce job.
* <p>
*
* @author <a href="mailto:thompsonbry@users.sourceforge.net">Bryan Thompson</a>
* @version $Id: ScaleOutTripleStore.java 3186 2010-07-12 17:01:19Z thompsonbry
* $
*/
public class ScaleOutTripleStore extends AbstractTripleStore {
/**
* Ctor specified by {@link DefaultResourceLocator}
* <p>
* Note: KB is NOT created automatically.
*/
public ScaleOutTripleStore(IIndexManager indexManager, String namespace,
Long timestamp, Properties properties) {
super( indexManager, namespace, timestamp, properties );
/*
* Note: The indexManager will be an IsolatedJournal if the resource is
* instantiated from within an AbstractTask.
*/
// this.fed = (IBigdataFederation) indexManager;
/*
* Note: KB is NOT created automatically.
*/
}
final public boolean isStable() {
return true;
// return getIndexManager().isStable();
}
// /**
// * Drops the indices for the {@link ITripleStore}.
// */
// public void __tearDownUnitTest() {
//
// destroy();
//
// super.__tearDownUnitTest();
//
// }
/**
* This store is safe for concurrent operations.
*/
public boolean isConcurrent() {
return true;
}
}