[jira] [Commented] (CARBONDATA-284) Abstracting Index and Segment interface

[Akash R Nilugal (Jira)]

    [ https://issues.apache.org/jira/browse/CARBONDATA-284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15612029#comment-15612029 ]

ASF GitHub Bot commented on CARBONDATA-284:
-------------------------------------------

Github user ravipesala commented on a diff in the pull request:

    https://github.com/apache/incubator-carbondata/pull/208#discussion_r85346673
 
    --- Diff: hadoop/src/main/java/org/apache/carbondata/hadoop/internal/segment/impl/IndexedSegment.java ---
    @@ -0,0 +1,73 @@
    +/*
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + *    http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing,
    + * software distributed under the License is distributed on an
    + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
    + * KIND, either express or implied.  See the License for the
    + * specific language governing permissions and limitations
    + * under the License.
    + */
    +
    +package org.apache.carbondata.hadoop.internal.segment.impl;
    +
    +import java.io.IOException;
    +import java.util.LinkedList;
    +import java.util.List;
    +
    +import org.apache.carbondata.hadoop.CarbonInputSplit;
    +import org.apache.carbondata.hadoop.internal.index.Block;
    +import org.apache.carbondata.hadoop.internal.segment.Segment;
    +import org.apache.carbondata.hadoop.internal.index.Index;
    +import org.apache.carbondata.hadoop.internal.index.IndexLoader;
    +import org.apache.carbondata.scan.filter.resolver.FilterResolverIntf;
    +import org.apache.hadoop.mapreduce.InputSplit;
    +import org.apache.hadoop.mapreduce.JobContext;
    +
    +/**
    + * This segment is backed by index, thus getSplits can use the index to do file pruning.
    + */
    +public class IndexedSegment extends Segment {
    +
    +  private IndexLoader loader;
    +
    +  public IndexedSegment(String name, String path, IndexLoader loader) {
    +    super(name, path);
    +    this.loader = loader;
    +  }
    +
    +  @Override
    +  public List<InputSplit> getSplits(JobContext job, FilterResolverIntf filterResolver)
    +      throws IOException {
    +    // do as following
    +    // 1. create the index or get from cache by the filter name in the configuration
    +    // 2. filter by index to get the filtered block
    +    // 3. create input split from filtered block
    +
    +    List<InputSplit> output = new LinkedList<>();
    +    Index index = loader.load(job.getConfiguration());
    --- End diff --
   
    does it required to load index every time?
    I guess we are just creating the instance of index here, so why don't you use factory here?

> Abstracting Index and Segment interface
> ---------------------------------------
>
>                 Key: CARBONDATA-284
>                 URL: https://issues.apache.org/jira/browse/CARBONDATA-284
>             Project: CarbonData
>          Issue Type: Improvement
>          Components: hadoop-integration
>    Affects Versions: 0.1.0-incubating
>            Reporter: Jacky Li
>             Fix For: 0.3.0-incubating
>
>
> This issue is intended to abstract developer API and user API to achieve following goals:
> Goal 1: User can choose the place to store Index data, it can be stored in
> processing framework's memory space (like in spark driver memory) or in
> another service outside of the processing framework (like using a
> independent database service, which can be shared across client)
> Goal 2: Developer can add more index of his choice to CarbonData files.
> Besides B+ tree on multi-dimensional key which current CarbonData supports,
> developers are free to add other indexing technology to make certain
> workload faster. These new indices should be added in a pluggable way.
> This Jira has been discussed in maillist:
> http://apache-carbondata-mailing-list-archive.1130556.n5.nabble.com/Abstracting-CarbonData-s-Index-Interface-td1587.html

--
This message was sent by Atlassian JIRA
(v6.3.4#6332)