Search in sources :

Example 1 with ProspectorServiceEvalStatsDAO

use of org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO in project incubator-rya by apache.

the class EntityOptimizer method setConf.

@Override
public void setConf(Configuration conf) {
    if (conf instanceof RdfCloudTripleStoreConfiguration) {
        this.conf = (RdfCloudTripleStoreConfiguration) conf;
    } else {
        this.conf = new AccumuloRdfConfiguration(conf);
    }
    if (!isEvalDaoSet) {
        if (this.conf.isUseStats() && this.conf.isUseSelectivity()) {
            try {
                eval = new AccumuloSelectivityEvalDAO(this.conf, ConfigUtils.getConnector(this.conf));
                ((AccumuloSelectivityEvalDAO) eval).setRdfEvalDAO(new ProspectorServiceEvalStatsDAO(ConfigUtils.getConnector(this.conf), this.conf));
                eval.init();
            } catch (final AccumuloException | AccumuloSecurityException e) {
                LOG.warn("A problem was encountered while setting the Configuration for the EntityOptimizer.", e);
            }
            isEvalDaoSet = true;
        } else {
            eval = null;
            isEvalDaoSet = true;
        }
    }
}
Also used : AccumuloException(org.apache.accumulo.core.client.AccumuloException) ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) AccumuloSelectivityEvalDAO(org.apache.rya.joinselect.AccumuloSelectivityEvalDAO) AccumuloSecurityException(org.apache.accumulo.core.client.AccumuloSecurityException) RdfCloudTripleStoreConfiguration(org.apache.rya.api.RdfCloudTripleStoreConfiguration) AccumuloRdfConfiguration(org.apache.rya.accumulo.AccumuloRdfConfiguration)

Example 2 with ProspectorServiceEvalStatsDAO

use of org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO in project incubator-rya by apache.

the class AccumuloSelectivityEvalDAOTest method init.

@Before
public void init() throws AccumuloException, AccumuloSecurityException, TableNotFoundException, TableExistsException {
    mock = new MockInstance("accumulo");
    PasswordToken pToken = new PasswordToken("pass".getBytes());
    conn = mock.getConnector("user", pToken);
    config = new BatchWriterConfig();
    config.setMaxMemory(1000);
    config.setMaxLatency(1000, TimeUnit.SECONDS);
    config.setMaxWriteThreads(10);
    if (conn.tableOperations().exists("rya_prospects")) {
        conn.tableOperations().delete("rya_prospects");
    }
    if (conn.tableOperations().exists("rya_selectivity")) {
        conn.tableOperations().delete("rya_selectivity");
    }
    arc = new AccumuloRdfConfiguration();
    res = new ProspectorServiceEvalStatsDAO(conn, arc);
    arc.setTableLayoutStrategy(new TablePrefixLayoutStrategy());
    arc.setMaxRangesForScanner(300);
}
Also used : PasswordToken(org.apache.accumulo.core.client.security.tokens.PasswordToken) MockInstance(org.apache.accumulo.core.client.mock.MockInstance) ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) TablePrefixLayoutStrategy(org.apache.rya.api.layout.TablePrefixLayoutStrategy) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) AccumuloRdfConfiguration(org.apache.rya.accumulo.AccumuloRdfConfiguration) Before(org.junit.Before)

Example 3 with ProspectorServiceEvalStatsDAO

use of org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO in project incubator-rya by apache.

the class RdfCloudTripleStoreSelectivityEvaluationStatisticsTest method testOptimizeQ1ZeroCard.

@Test
public void testOptimizeQ1ZeroCard() throws Exception {
    RdfEvalStatsDAO<RdfCloudTripleStoreConfiguration> res = new ProspectorServiceEvalStatsDAO(conn, arc);
    AccumuloSelectivityEvalDAO accc = new AccumuloSelectivityEvalDAO();
    accc.setConf(arc);
    accc.setConnector(conn);
    accc.setRdfEvalDAO(res);
    accc.init();
    BatchWriter bw1 = conn.createBatchWriter("rya_prospects", config);
    BatchWriter bw2 = conn.createBatchWriter("rya_selectivity", config);
    String s1 = "predicateobject" + DELIM + "http://www.w3.org/2000/01/rdf-schema#label" + DELIM + "uri:dog";
    String s2 = "predicateobject" + DELIM + "uri:barksAt" + DELIM + "uri:cat";
    String s3 = "predicateobject" + DELIM + "uri:peesOn" + DELIM + "uri:hydrant";
    List<Mutation> mList = new ArrayList<Mutation>();
    List<Mutation> mList2 = new ArrayList<Mutation>();
    List<String> sList = Arrays.asList("subjectobject", "subjectpredicate", "subjectsubject", "predicateobject", "predicatepredicate", "predicatesubject");
    Mutation m1, m2, m3, m4;
    m1 = new Mutation(s1 + DELIM + "1");
    m1.put(new Text("count"), new Text(""), new Value("1".getBytes()));
    m2 = new Mutation(s2 + DELIM + "2");
    m2.put(new Text("count"), new Text(""), new Value("2".getBytes()));
    // m3 = new Mutation(s3 + DELIM + "3");
    // m3.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    mList.add(m1);
    mList.add(m2);
    // mList.add(m3);
    bw1.addMutations(mList);
    bw1.close();
    // Scanner scan = conn.createScanner("rya_prospects", new Authorizations());
    // scan.setRange(new Range());
    // for (Map.Entry<Key, Value> entry : scan) {
    // System.out.println("Key row string is " + entry.getKey().getRow().toString());
    // System.out.println("Key is " + entry.getKey());
    // System.out.println("Value is " + (new String(entry.getValue().get())));
    // }
    m1 = new Mutation(s1);
    m2 = new Mutation(s2);
    m3 = new Mutation(s3);
    m4 = new Mutation(new Text("subjectpredicateobject" + DELIM + "FullTableCardinality"));
    m4.put(new Text("FullTableCardinality"), new Text("100"), EMPTY_VAL);
    int i = 2;
    int j = 3;
    int k = 4;
    Long count1;
    Long count2;
    Long count3;
    for (String s : sList) {
        count1 = (long) i;
        count2 = (long) j;
        count3 = (long) k;
        m1.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m2.put(new Text(s), new Text(count2.toString()), EMPTY_VAL);
        m3.put(new Text(s), new Text(count3.toString()), EMPTY_VAL);
        i = 2 * i;
        j = 2 * j;
        k = 2 * k;
    }
    mList2.add(m1);
    mList2.add(m2);
    mList2.add(m3);
    mList2.add(m4);
    bw2.addMutations(mList2);
    bw2.close();
    // scan = conn.createScanner("rya_selectivity", new Authorizations());
    // scan.setRange(new Range());
    // for (Map.Entry<Key, Value> entry : scan) {
    // System.out.println("Key row string is " + entry.getKey().getRow().toString());
    // System.out.println("Key is " + entry.getKey());
    // System.out.println("Value is " + (new String(entry.getKey().getColumnQualifier().toString())));
    // 
    // }
    TupleExpr te = getTupleExpr(q1);
    System.out.println(te);
    RdfCloudTripleStoreSelectivityEvaluationStatistics ars = new RdfCloudTripleStoreSelectivityEvaluationStatistics(arc, res, accc);
    double card = ars.getCardinality(te);
    Assert.assertEquals(4.04, card, .0001);
}
Also used : ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) ArrayList(java.util.ArrayList) AccumuloSelectivityEvalDAO(org.apache.rya.joinselect.AccumuloSelectivityEvalDAO) Text(org.apache.hadoop.io.Text) TupleExpr(org.openrdf.query.algebra.TupleExpr) Value(org.apache.accumulo.core.data.Value) BatchWriter(org.apache.accumulo.core.client.BatchWriter) Mutation(org.apache.accumulo.core.data.Mutation) RdfCloudTripleStoreConfiguration(org.apache.rya.api.RdfCloudTripleStoreConfiguration) Test(org.junit.Test)

Example 4 with ProspectorServiceEvalStatsDAO

use of org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO in project incubator-rya by apache.

the class QueryJoinSelectOptimizerTest method testOptimizeQ3.

@Test
public void testOptimizeQ3() throws Exception {
    RdfEvalStatsDAO<RdfCloudTripleStoreConfiguration> res = new ProspectorServiceEvalStatsDAO(conn, arc);
    AccumuloSelectivityEvalDAO accc = new AccumuloSelectivityEvalDAO();
    accc.setConf(arc);
    accc.setConnector(conn);
    accc.setRdfEvalDAO(res);
    accc.init();
    BatchWriter bw1 = conn.createBatchWriter("rya_prospects", config);
    BatchWriter bw2 = conn.createBatchWriter("rya_selectivity", config);
    String s1 = "predicateobject" + DELIM + "http://www.w3.org/2000/01/rdf-schema#label" + DELIM + "uri:dog";
    String s2 = "predicateobject" + DELIM + "uri:barksAt" + DELIM + "uri:cat";
    String s3 = "predicateobject" + DELIM + "uri:peesOn" + DELIM + "uri:hydrant";
    String s5 = "predicateobject" + DELIM + "uri:scratches" + DELIM + "uri:ears";
    String s4 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:chickens";
    String s6 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:kibble";
    String s7 = "predicateobject" + DELIM + "uri:rollsIn" + DELIM + "uri:mud";
    String s8 = "predicateobject" + DELIM + "uri:runsIn" + DELIM + "uri:field";
    String s9 = "predicateobject" + DELIM + "uri:smells" + DELIM + "uri:butt";
    String s10 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:sticks";
    List<Mutation> mList = new ArrayList<Mutation>();
    List<Mutation> mList2 = new ArrayList<Mutation>();
    List<String> sList = Arrays.asList("subjectobject", "subjectpredicate", "subjectsubject", "predicateobject", "predicatepredicate", "predicatesubject");
    Mutation m1, m2, m3, m4, m5, m6, m7, m8, m9, m10, m11;
    m1 = new Mutation(s1 + DELIM + "3");
    m1.put(new Text("count"), new Text(""), new Value("5".getBytes()));
    m2 = new Mutation(s2 + DELIM + "2");
    m2.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m3 = new Mutation(s3 + DELIM + "1");
    m3.put(new Text("count"), new Text(""), new Value("2".getBytes()));
    m4 = new Mutation(s4 + DELIM + "1");
    m4.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m5 = new Mutation(s5 + DELIM + "1");
    m5.put(new Text("count"), new Text(""), new Value("5".getBytes()));
    m6 = new Mutation(s6 + DELIM + "1");
    m6.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m7 = new Mutation(s7 + DELIM + "1");
    m7.put(new Text("count"), new Text(""), new Value("2".getBytes()));
    m8 = new Mutation(s8 + DELIM + "1");
    m8.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m9 = new Mutation(s9 + DELIM + "1");
    m9.put(new Text("count"), new Text(""), new Value("1".getBytes()));
    m10 = new Mutation(s10 + DELIM + "1");
    m10.put(new Text("count"), new Text(""), new Value("1".getBytes()));
    mList.add(m1);
    mList.add(m2);
    mList.add(m3);
    mList.add(m4);
    mList.add(m5);
    mList.add(m6);
    mList.add(m7);
    mList.add(m8);
    mList.add(m9);
    mList.add(m10);
    bw1.addMutations(mList);
    bw1.close();
    Scanner scan = conn.createScanner("rya_prospects", new Authorizations());
    scan.setRange(new Range());
    for (Map.Entry<Key, Value> entry : scan) {
        System.out.println("Key row string is " + entry.getKey().getRow().toString());
        System.out.println("Key is " + entry.getKey());
        System.out.println("Value is " + (new String(entry.getValue().get())));
    }
    m1 = new Mutation(s1);
    m2 = new Mutation(s2);
    m3 = new Mutation(s3);
    m4 = new Mutation(s4);
    m5 = new Mutation(s5);
    m6 = new Mutation(s6);
    m7 = new Mutation(s7);
    m8 = new Mutation(s8);
    m9 = new Mutation(s9);
    m10 = new Mutation(s10);
    m11 = new Mutation(new Text("subjectpredicateobject" + DELIM + "FullTableCardinality"));
    m11.put(new Text("FullTableCardinality"), new Text("100"), EMPTY_VAL);
    int i = 2;
    int j = 3;
    int k = 4;
    int l = 5;
    Long count1;
    Long count2;
    Long count3;
    Long count4;
    for (String s : sList) {
        count1 = (long) i;
        count2 = (long) j;
        count3 = (long) k;
        count4 = (long) l;
        m1.put(new Text(s), new Text(count4.toString()), EMPTY_VAL);
        m2.put(new Text(s), new Text(count2.toString()), EMPTY_VAL);
        m3.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m4.put(new Text(s), new Text(count3.toString()), EMPTY_VAL);
        m5.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m6.put(new Text(s), new Text(count2.toString()), EMPTY_VAL);
        m7.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m8.put(new Text(s), new Text(count4.toString()), EMPTY_VAL);
        m9.put(new Text(s), new Text(count3.toString()), EMPTY_VAL);
        m10.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        i = 2 * i;
        j = 2 * j;
        k = 2 * k;
        l = 2 * l;
    }
    mList2.add(m1);
    mList2.add(m2);
    mList2.add(m3);
    mList2.add(m5);
    mList2.add(m4);
    mList2.add(m6);
    mList2.add(m7);
    mList2.add(m8);
    mList2.add(m9);
    mList2.add(m10);
    mList2.add(m11);
    bw2.addMutations(mList2);
    bw2.close();
    scan = conn.createScanner("rya_selectivity", new Authorizations());
    scan.setRange(new Range());
    for (Map.Entry<Key, Value> entry : scan) {
        System.out.println("Key row string is " + entry.getKey().getRow().toString());
        System.out.println("Key is " + entry.getKey());
        System.out.println("Value is " + (new String(entry.getKey().getColumnQualifier().toString())));
    }
    TupleExpr te = getTupleExpr(q3);
    RdfCloudTripleStoreSelectivityEvaluationStatistics ars = new RdfCloudTripleStoreSelectivityEvaluationStatistics(arc, res, accc);
    QueryJoinSelectOptimizer qjs = new QueryJoinSelectOptimizer(ars, accc);
    System.out.println("Originial query is " + te);
    qjs.optimize(te, null, null);
    System.out.print("Optimized query is " + te);
}
Also used : Scanner(org.apache.accumulo.core.client.Scanner) Authorizations(org.apache.accumulo.core.security.Authorizations) ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) ArrayList(java.util.ArrayList) AccumuloSelectivityEvalDAO(org.apache.rya.joinselect.AccumuloSelectivityEvalDAO) QueryJoinSelectOptimizer(org.apache.rya.rdftriplestore.evaluation.QueryJoinSelectOptimizer) Text(org.apache.hadoop.io.Text) RdfCloudTripleStoreSelectivityEvaluationStatistics(org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics) Range(org.apache.accumulo.core.data.Range) TupleExpr(org.openrdf.query.algebra.TupleExpr) Value(org.apache.accumulo.core.data.Value) BatchWriter(org.apache.accumulo.core.client.BatchWriter) Mutation(org.apache.accumulo.core.data.Mutation) RdfCloudTripleStoreConfiguration(org.apache.rya.api.RdfCloudTripleStoreConfiguration) Map(java.util.Map) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Example 5 with ProspectorServiceEvalStatsDAO

use of org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO in project incubator-rya by apache.

the class QueryJoinSelectOptimizerTest method init.

@Before
public void init() throws AccumuloException, AccumuloSecurityException, TableNotFoundException, TableExistsException {
    mock = new MockInstance("accumulo");
    PasswordToken pToken = new PasswordToken("pass".getBytes());
    conn = mock.getConnector("user", pToken);
    config = new BatchWriterConfig();
    config.setMaxMemory(1000);
    config.setMaxLatency(1000, TimeUnit.SECONDS);
    config.setMaxWriteThreads(10);
    if (conn.tableOperations().exists("rya_prospects")) {
        conn.tableOperations().delete("rya_prospects");
    }
    if (conn.tableOperations().exists("rya_selectivity")) {
        conn.tableOperations().delete("rya_selectivity");
    }
    arc = new AccumuloRdfConfiguration();
    arc.setTableLayoutStrategy(new TablePrefixLayoutStrategy());
    arc.setMaxRangesForScanner(300);
    res = new ProspectorServiceEvalStatsDAO(conn, arc);
}
Also used : PasswordToken(org.apache.accumulo.core.client.security.tokens.PasswordToken) MockInstance(org.apache.accumulo.core.client.mock.MockInstance) TablePrefixLayoutStrategy(org.apache.rya.api.layout.TablePrefixLayoutStrategy) ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) AccumuloRdfConfiguration(org.apache.rya.accumulo.AccumuloRdfConfiguration) Before(org.junit.Before)

Aggregations

ProspectorServiceEvalStatsDAO (org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO)14 RdfCloudTripleStoreConfiguration (org.apache.rya.api.RdfCloudTripleStoreConfiguration)10 AccumuloSelectivityEvalDAO (org.apache.rya.joinselect.AccumuloSelectivityEvalDAO)10 ArrayList (java.util.ArrayList)8 BatchWriter (org.apache.accumulo.core.client.BatchWriter)8 Mutation (org.apache.accumulo.core.data.Mutation)8 Value (org.apache.accumulo.core.data.Value)8 Text (org.apache.hadoop.io.Text)8 Test (org.junit.Test)8 TupleExpr (org.openrdf.query.algebra.TupleExpr)8 Map (java.util.Map)6 Scanner (org.apache.accumulo.core.client.Scanner)6 Key (org.apache.accumulo.core.data.Key)6 Range (org.apache.accumulo.core.data.Range)6 Authorizations (org.apache.accumulo.core.security.Authorizations)6 QueryJoinSelectOptimizer (org.apache.rya.rdftriplestore.evaluation.QueryJoinSelectOptimizer)6 RdfCloudTripleStoreSelectivityEvaluationStatistics (org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics)6 AccumuloRdfConfiguration (org.apache.rya.accumulo.AccumuloRdfConfiguration)5 BatchWriterConfig (org.apache.accumulo.core.client.BatchWriterConfig)3 MockInstance (org.apache.accumulo.core.client.mock.MockInstance)3