HBase中加盐（Salting）之后的表如何读取：协处理器文章

我们介绍了避免数据斑点的三种比较常见方法：

• 加盐-盐腌
• 哈希-散列
• 反转-反转

其中在加盐（Salting）的方法里面是这么描述的：给Rowkey分配一个随机指针以使其和之前排序不同。但是在Rowkey前面加了随机重叠，那么我们怎么将这些数据替换来呢？我将分三篇文章来介绍如何读取加盐之后的表，其中每篇文章提供一种方法，主要包括：

• 使用协处理器读取加盐的表
• 使用Spark读取加盐的表
• 使用MapReduce读取加盐的表

关于协处理器的入门及实战，参见请这里。本文使用的各组件版本：Hadoop的2.7.7，HBase的-2.0.4，jdk1.8.0_201。

测试数据生成

在介绍如何查询数据之前，我们先创建一张名为iteblog的HBase表，进行测试。为了数据均匀和介绍的方便，这里使用了预分区，并设置了27个分区，如下：

hbase(main):002:0> create 'iteblog', 'f', SPLITS => ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z']

0 row(s) in 2.4880 seconds

然后我们使用以下方法生成了1000000条测试数据。RowKey的形式为UID +当前数据生成版本；由于UID的长度为4，所以1000000条数据会存在大量的UID一样的数据，所以我们使用加盐方法将这些数据均匀分散到上述27个Region里面（注意，其实第一个Region其实没数据）。具体代码如下：

package com.iteblog.data;

 

import org.apache.hadoop.conf.Configuration;

import org.apache.hadoop.hbase.HBaseConfiguration;

import org.apache.hadoop.hbase.HConstants;

import org.apache.hadoop.hbase.TableName;

import org.apache.hadoop.hbase.client.*;

import org.apache.hadoop.hbase.util.Bytes;

 

import java.io.IOException;

import java.util.ArrayList;

import java.util.List;

import java.util.Random;

import java.util.UUID;

 

public class HBaseDataGenerator {

    private static byte[] FAMILY = "f".getBytes();

    private static byte[] QUALIFIER_UUID = "uuid".getBytes();

    private static byte[] QUALIFIER_AGE = "age".getBytes();

 

    private static char generateLetter() {

        return (char) (Math.random() * 26 + 'A');

    }

 

    private static long generateUid(int n) {

        return (long) (Math.random() * 9 * Math.pow(10, n - 1)) + (long) Math.pow(10, n - 1);

    }

 

    public static void main(String[] args) throws IOException {

        BufferedMutatorParams bmp = new BufferedMutatorParams(TableName.valueOf("iteblog"));

        bmp.writeBufferSize(1024 * 1024 * 24);

 

        Configuration conf = HBaseConfiguration.create();

        conf.set(HConstants.ZOOKEEPER_QUORUM, "https://www.iteblog.com:2181");

        Connection connection = ConnectionFactory.createConnection(conf);

 

        BufferedMutator bufferedMutator = connection.getBufferedMutator(bmp);

 

        int BATCH_SIZE = 1000;

        int COUNTS = 1000000;

        int count = 0;

        List<Put> putList = new ArrayList<>();

 

        for (int i = 0; i < COUNTS; i++) {

            String rowKey = generateLetter() + "-"

                    + generateUid(4) + "-"

                    + System.currentTimeMillis();

 

            Put put = new Put(Bytes.toBytes(rowKey));

            byte[] uuidBytes = UUID.randomUUID().toString().substring(0, 23).getBytes();

            put.addColumn(FAMILY, QUALIFIER_UUID, uuidBytes);

            put.addColumn(FAMILY, QUALIFIER_AGE, Bytes.toBytes("" + new Random().nextInt(100)));

            putList.add(put);

            count++;

 

            if (count % BATCH_SIZE == 0) {

                bufferedMutator.mutate(putList);

                bufferedMutator.flush();

                putList.clear();

                System.out.println(count);

            }

        }

 

        if (putList.size() > 0) {

            bufferedMutator.mutate(putList);

            bufferedMutator.flush();

            putList.clear();

        }

 

    }

}

运行完上面的代码之后，会生成1000000条数据（注意，这里其实不严谨，因为Rowkey设计问题，可能会导致重复的Rowkey生成，所以实际情况下可能没有1000000条数据。）。我们limit 10条数据看下长成什么样：

hbase(main):001:0> scan 'iteblog', {'LIMIT'=>10}

ROW                        COLUMN+CELL

 A-1000-1550572395399      column=f:age, timestamp=1549091990253, value=54

 A-1000-1550572395399      column=f:uuid, timestamp=1549091990253, value=e9b10a9f-1218-43fd-bd01

 A-1000-1550572413799      column=f:age, timestamp=1549092008575, value=4

 A-1000-1550572413799      column=f:uuid, timestamp=1549092008575, value=181aa91e-5f1d-454c-959c

 A-1000-1550572414761      column=f:age, timestamp=1549092009531, value=33

 A-1000-1550572414761      column=f:uuid, timestamp=1549092009531, value=19aad8d3-621a-473c-8f9f

 A-1001-1550572394570      column=f:age, timestamp=1549091989341, value=64

 A-1001-1550572394570      column=f:uuid, timestamp=1549091989341, value=c6712a0d-3793-46d5-865b

 A-1001-1550572405337      column=f:age, timestamp=1549092000108, value=96

 A-1001-1550572405337      column=f:uuid, timestamp=1549092000108, value=4bf05d10-bb4d-43e3-9957

 A-1001-1550572419688      column=f:age, timestamp=1549092014458, value=8

 A-1001-1550572419688      column=f:uuid, timestamp=1549092014458, value=f04ba835-d8ac-49a3-8f96

 A-1002-1550572424041      column=f:age, timestamp=1549092018816, value=84

 A-1002-1550572424041      column=f:uuid, timestamp=1549092018816, value=99d6c989-afb5-4101-9d95

 A-1003-1550572431830      column=f:age, timestamp=1549092026605, value=21

 A-1003-1550572431830      column=f:uuid, timestamp=1549092026605, value=8c1ff1b6-b97c-4059-9b68

 A-1004-1550572395399      column=f:age, timestamp=1549091990253, value=2

 A-1004-1550572395399      column=f:uuid, timestamp=1549091990253, value=e240aa0f-c044-452f-89c0

 A-1004-1550572403783      column=f:age, timestamp=1549091998555, value=6

 A-1004-1550572403783      column=f:uuid, timestamp=1549091998555, value=e8df15c9-02fa-458e-bd0c

10 row(s)

Took 0.1104 seconds

使用协处理器查询加盐之后的表

现在有数据了，我们需要查询所有UID = 1000的用户所有历史数据，那么如何查呢？我们知道UID = 1000的用户数据是均匀放到上述的27个地区里面的，因为经过加盐了，所以这些数据垂直都是垂直A-，B-，C-等开头的。其次我们需要知道，每个区域其实是有Start Key和End Key的，这些Start Key和End Key其实就是我们创建iteblog表指定的。如果你看了《 HBase协处理器入门及实战》这篇文章，你就知道协处理器的代码其实是在每个区域里面执行的；而这些代码在区域里面执行的时候是可以拿到当前Region的信息，包括了键和结束键，所以实际上我们可以将拿到的开始键信息和查询的UID进行拆分，这样就可以查询我们要的数据。协处理器处理文章就是基于这样的思想来查询加盐之后的数据的。

定义proto文件

为什么需要定义这个请参见《 HBase协处理器入门及实战》这篇文章。因为我们查询的时候需要引用查询的参数，表名，StartKey，EndKey以及是否加盐等标记；同时当查询到结果的当时，我们还需要将数据返回，所以我们定义的proto文件如下：

option java_package = "com.iteblog.data.coprocessor.generated";

option java_outer_classname = "DataQueryProtos";

option java_generic_services = true;

option java_generate_equals_and_hash = true;

option optimize_for = SPEED;

 

message DataQueryRequest {

  optional string tableName = 1;

  optional string startRow = 2;

  optional string endRow = 3;

  optional bool  incluedEnd = 4;

  optional bool  isSalting = 5;

}

 

message DataQueryResponse {

  message Cell{

    required bytes value = 1;

    required bytes family = 2;

    required bytes qualifier = 3;

    required bytes row = 4;

    required int64 timestamp = 5;

  }

 

  message Row{

    optional bytes rowKey = 1;

    repeated Cell cellList = 2;

  }

 

  repeated Row rowList = 1;

}

 

service QueryDataService{

  rpc queryByStartRowAndEndRow(DataQueryRequest)

    returns (DataQueryResponse);

}

我们然后使用protobuf-maven-plugin插件将上面的原生成的Java类，具体如何操作参见“在IDEA中使用Maven的编译原文件”。将我们的生成DataQueryProtos.java类拷贝产品到com.iteblog.data.coprocessor.generated包里面。

编写协处理器代码

有了请求和返回的类，现在我们需要编写协处理器的处理代码了，结合上面的分析，协处理器的代码实现如下：

package com.iteblog.data.coprocessor;

 

import com.google.protobuf.ByteString;

import com.google.protobuf.RpcCallback;

import com.google.protobuf.RpcController;

import com.google.protobuf.Service;

import com.iteblog.data.coprocessor.generated.DataQueryProtos.QueryDataService;

import com.iteblog.data.coprocessor.generated.DataQueryProtos.DataQueryRequest;

import com.iteblog.data.coprocessor.generated.DataQueryProtos.DataQueryResponse;

import org.apache.hadoop.hbase.Cell;

import org.apache.hadoop.hbase.CoprocessorEnvironment;

import org.apache.hadoop.hbase.client.Get;

import org.apache.hadoop.hbase.client.Result;

import org.apache.hadoop.hbase.client.Scan;

import org.apache.hadoop.hbase.coprocessor.CoprocessorException;

import org.apache.hadoop.hbase.coprocessor.RegionCoprocessor;

import org.apache.hadoop.hbase.coprocessor.RegionCoprocessorEnvironment;

import org.apache.hadoop.hbase.regionserver.InternalScanner;

import org.apache.hadoop.hbase.shaded.protobuf.ResponseConverter;

import org.apache.hadoop.hbase.util.Bytes;

 

import java.io.IOException;

import java.util.ArrayList;

import java.util.Collections;

import java.util.List;

 

public class SlatTableDataSearch extends QueryDataService implements RegionCoprocessor {

    private RegionCoprocessorEnvironment env;

 

    public Iterable<Service> getServices() {

        return Collections.singleton(this);

    }

 

    @Override

    public void queryByStartRowAndEndRow(RpcController controller,

                                         DataQueryRequest request,

                                         RpcCallback<DataQueryResponse> done) {

        DataQueryResponse response = null;

 

        String startRow = request.getStartRow();

        String endRow = request.getEndRow();

        String regionStartKey = Bytes.toString(this.env.getRegion().getRegionInfo().getStartKey());

 

        if (request.getIsSalting()) {

            String startSalt = null;

            if (null != regionStartKey && regionStartKey.length() != 0) {

                startSalt = regionStartKey;

            }

            if (null != startSalt && null != startRow) {

                startRow = startSalt + "-" + startRow;

                endRow = startSalt + "-" + endRow;

            }

        }

 

        Scan scan = new Scan();

        if (null != startRow) {

            scan.withStartRow(Bytes.toBytes(startRow));

        }

 

        if (null != endRow) {

            scan.withStopRow(Bytes.toBytes(endRow), request.getIncluedEnd());

        }

 

        try (InternalScanner scanner = this.env.getRegion().getScanner(scan)) {

            List<Cell> results = new ArrayList<>();

 

            boolean hasMore;

            DataQueryResponse.Builder responseBuilder = DataQueryResponse.newBuilder();

            do {

                hasMore = scanner.next(results);

                DataQueryResponse.Row.Builder rowBuilder = DataQueryResponse.Row.newBuilder();

                if (results.size() > 0) {

                    Cell cell = results.get(0);

                    rowBuilder.setRowKey(ByteString.copyFrom(cell.getRowArray(), cell.getRowOffset(), cell.getRowLength()));

                    for (Cell kv : results) {

                        buildCell(rowBuilder, kv);

                    }

                }

 

                responseBuilder.addRowList(rowBuilder);

                results.clear();

            } while (hasMore);

 

            response = responseBuilder.build();

 

        } catch (IOException e) {

            ResponseConverter.setControllerException(controller, e);

        }

        done.run(response);

    }

 

    private void buildCell(DataQueryResponse.Row.Builder rowBuilder, Cell kv) {

        DataQueryResponse.Cell.Builder cellBuilder = DataQueryResponse.Cell.newBuilder();

        cellBuilder.setFamily(ByteString.copyFrom(kv.getFamilyArray(), kv.getFamilyOffset(), kv.getFamilyLength()));

        cellBuilder.setQualifier(ByteString.copyFrom(kv.getQualifierArray(), kv.getQualifierOffset(), kv.getQualifierLength()));

        cellBuilder.setRow(ByteString.copyFrom(kv.getRowArray(), kv.getRowOffset(), kv.getRowLength()));

        cellBuilder.setValue(ByteString.copyFrom(kv.getValueArray(), kv.getValueOffset(), kv.getValueLength()));

        cellBuilder.setTimestamp(kv.getTimestamp());

        rowBuilder.addCellList(cellBuilder);

    }

 

    /**

     * Stores a reference to the coprocessor environment provided by the

     * {@link org.apache.hadoop.hbase.regionserver.RegionCoprocessorHost} from the region where this

     * coprocessor is loaded.  Since this is a coprocessor endpoint, it always expects to be loaded

     * on a table region, so always expects this to be an instance of

     * {@link RegionCoprocessorEnvironment}.

     *

     * @param env the environment provided by the coprocessor host

     * @throws IOException if the provided environment is not an instance of

     *                     {@code RegionCoprocessorEnvironment}

     */

    @Override

    public void start(CoprocessorEnvironment env) throws IOException {

        if (env instanceof RegionCoprocessorEnvironment) {

            this.env = (RegionCoprocessorEnvironment) env;

        } else {

            throw new CoprocessorException("Must be loaded on a table region!");

        }

    }

 

    @Override

    public void stop(CoprocessorEnvironment env) {

        // nothing to do

    }

}

大家可以看到，这里面的代码框架和《 HBase协处理器入门及实战》里面介绍的HBase提供的RowCountEndpoint示例代码很类似。主要逻辑在queryByStartRowAndEndRow函数实现里面。我们通过DataQueryRequest拿到客户端查询的表，StartKey和EndKey等数据。通过this.env.getRegion().getRegionInfo().getStartKey()可以拿到当前区域的StartKey，然后再和和客户端传进来的StartKey和EndKey进行拼接就可以拿到完整的Rowkey插入。剩下的查询就是正常的HBase扫描代码了。

现在我们将SlatTableDataSearch类进行编译打包，并部署到HBase表里面去，具体如何部署参见《 HBase协处理器入门及实战》

协处理器客户端代码编写

到这里，我们的协处理器服务器端的代码和部署已经完成了，现在我们需要编写协处理器客户端代码。其实也很简单，如下：

package com.iteblog.data;

 

import com.iteblog.data.coprocessor.generated.DataQueryProtos.QueryDataService;

import com.iteblog.data.coprocessor.generated.DataQueryProtos.DataQueryRequest;

import com.iteblog.data.coprocessor.generated.DataQueryProtos.DataQueryResponse;

import com.iteblog.data.coprocessor.generated.DataQueryProtos.DataQueryResponse.*;

import org.apache.hadoop.conf.Configuration;

import org.apache.hadoop.hbase.HBaseConfiguration;

import org.apache.hadoop.hbase.TableName;

import org.apache.hadoop.hbase.client.Connection;

import org.apache.hadoop.hbase.client.ConnectionFactory;

import org.apache.hadoop.hbase.client.HTable;

import org.apache.hadoop.hbase.ipc.CoprocessorRpcUtils.BlockingRpcCallback;

import org.apache.hadoop.hbase.ipc.ServerRpcController;

 

import java.util.LinkedList;

import java.util.List;

import java.util.Map;

 

public class DataQuery {

    private static Configuration conf = null;

 

    static {

        conf = HBaseConfiguration.create();

        conf.set("hbase.zookeeper.quorum", "https://www.iteblog.com:2181");

    }

 

    static List<Row> queryByStartRowAndStopRow(String tableName,

                                               String startRow, String stopRow,

                                               boolean isIncludeEnd, boolean isSalting) {

 

        final DataQueryRequest.Builder requestBuilder = DataQueryRequest.newBuilder();

        requestBuilder.setTableName(tableName);

        requestBuilder.setStartRow(startRow);

        requestBuilder.setEndRow(stopRow);

        requestBuilder.setIncluedEnd(isIncludeEnd);

        requestBuilder.setIsSalting(isSalting);

 

        try {

            Connection connection = ConnectionFactory.createConnection(conf);

            HTable table = (HTable) connection.getTable(TableName.valueOf(tableName));

            Map<byte[], List<Row>> result = table.coprocessorService(QueryDataService.class,

                    null, null, counter -> {

                        ServerRpcController controller = new ServerRpcController();

                        BlockingRpcCallback<DataQueryResponse> call = new BlockingRpcCallback<>();

                        counter.queryByStartRowAndEndRow(controller, requestBuilder.build(), call);

                        DataQueryResponse response = call.get();

 

                        if (controller.failedOnException()) {

                            throw controller.getFailedOn();

                        }

 

                        return response.getRowListList();

                    });

 

            List<Row> list = new LinkedList<>();

            for (Map.Entry<byte[], List<Row>> entry : result.entrySet()) {

                if (null != entry.getKey()) {

                    list.addAll(entry.getValue());

                }

            }

            return list;

        } catch (Throwable e) {

            e.printStackTrace();

        }

        return null;

 

    }

 

    public static void main(String[] args) {

        List<Row> rows = queryByStartRowAndStopRow("iteblog", "1000", "1001", false, true);

        if (null != rows) {

            System.out.println(rows.size());

            for (DataQueryResponse.Row row : rows) {

                List<DataQueryResponse.Cell> cellListList = row.getCellListList();

                for (DataQueryResponse.Cell cell : cellListList) {

                    System.out.println(row.getRowKey().toStringUtf8() + " " +

                            "column=" + cell.getFamily().toStringUtf8() +

                            ":" + cell.getQualifier().toStringUtf8() + ", " +

                            "timestamp=" + cell.getTimestamp() + ", " +

                            "value=" + cell.getValue().toStringUtf8());

                }

            }

        }

    }

}

我们运行上面的代码，可以得到如下的输出：

A-1000-1550572395399     column=f:age, timestamp=1549091990253, value=54

A-1000-1550572395399     column=f:uuid, timestamp=1549091990253, value=e9b10a9f-1218-43fd-bd01

A-1000-1550572413799     column=f:age, timestamp=1549092008575, value=4

A-1000-1550572413799     column=f:uuid, timestamp=1549092008575, value=181aa91e-5f1d-454c-959c

A-1000-1550572414761     column=f:age, timestamp=1549092009531, value=33

A-1000-1550572414761     column=f:uuid, timestamp=1549092009531, value=19aad8d3-621a-473c-8f9f

B-1000-1550572388491     column=f:age, timestamp=1549091983276, value=1

B-1000-1550572388491     column=f:uuid, timestamp=1549091983276, value=cf720efe-2ad2-48d6-81b8

B-1000-1550572392922     column=f:age, timestamp=1549091987701, value=7

B-1000-1550572392922     column=f:uuid, timestamp=1549091987701, value=8a047118-e130-48cb-adfe

 

hbase(main):020:0> scan 'iteblog', {STARTROW => 'A-1000', ENDROW => 'A-1001'}

ROW                         COLUMN+CELL

 A-1000-1550572395399       column=f:age, timestamp=1549091990253, value=54

 A-1000-1550572395399       column=f:uuid, timestamp=1549091990253, value=e9b10a9f-1218-43fd-bd01

 A-1000-1550572413799       column=f:age, timestamp=1549092008575, value=4

 A-1000-1550572413799       column=f:uuid, timestamp=1549092008575, value=181aa91e-5f1d-454c-959c

 A-1000-1550572414761       column=f:age, timestamp=1549092009531, value=33

 A-1000-1550572414761       column=f:uuid, timestamp=1549092009531, value=19aad8d3-621a-473c-8f9f

3 row(s)

Took 0.0569 seconds

可以看到，和我们使用HBase Shell输出的一致，而且我们还把所有的UID = 1000的数据拿到了。好了，到这里，使用协处理器查询HBase加盐之后的表已经算完成了，明天我将介绍使用Spark如何查询加盐之后的表。