高并发导入优化（Group Commit）

CREATE TABLE `dt` (
    `id` int(11) NOT NULL,
    `name` varchar(50) NULL,
    `score` int(11) NULL
) ENGINE=OLAP
DUPLICATE KEY(`id`)
DISTRIBUTED BY HASH(`id`) BUCKETS 1
PROPERTIES (
    "replication_num" = "1"
);

url = jdbc:mysql://127.0.0.1:9030/db?useServerPrepStmts=true&useLocalSessionState=true&rewriteBatchedStatements=true&cachePrepStmts=true&prepStmtCacheSqlLimit=99999&prepStmtCacheSize=500

url = jdbc:mysql://127.0.0.1:9030/db?useServerPrepStmts=true&useLocalSessionState=true&rewriteBatchedStatements=true&cachePrepStmts=true&prepStmtCacheSqlLimit=99999&prepStmtCacheSize=500&sessionVariables=group_commit=async_mode

try (Statement statement = conn.createStatement()) {
    statement.execute("SET group_commit = async_mode;");
}

private static final String JDBC_DRIVER = "com.mysql.jdbc.Driver";
private static final String URL_PATTERN = "jdbc:mysql://%s:%d/%s?useServerPrepStmts=true&useLocalSessionState=true&rewriteBatchedStatements=true&cachePrepStmts=true&prepStmtCacheSqlLimit=99999&prepStmtCacheSize=50&sessionVariables=group_commit=async_mode";
private static final String HOST = "127.0.0.1";
private static final int PORT = 9087;
private static final String DB = "db";
private static final String TBL = "dt";
private static final String USER = "root";
private static final String PASSWD = "";
private static final int INSERT_BATCH_SIZE = 10;

private static void groupCommitInsertBatch() throws Exception {
    Class.forName(JDBC_DRIVER);
    // add rewriteBatchedStatements=true and cachePrepStmts=true in JDBC url
    // set session variables by sessionVariables=group_commit=async_mode in JDBC url
    try (Connection conn = DriverManager.getConnection(
            String.format(URL_PATTERN, HOST, PORT, DB), USER, PASSWD)) {

        String query = "insert into " + TBL + " values(?, ?, ?)";
        try (PreparedStatement stmt = conn.prepareStatement(query)) {
            for (int j = 0; j < 5; j++) {
                // 10 rows per insert
                for (int i = 0; i < INSERT_BATCH_SIZE; i++) {
                    stmt.setInt(1, i);
                    stmt.setString(2, "name" + i);
                    stmt.setInt(3, i + 10);
                    stmt.addBatch();
                }
                int[] result = stmt.executeBatch();
            }
        }
    } catch (Exception e) {
        e.printStackTrace();
    }
}

# 配置 session 变量开启打印parpared语句的audit log, 默认为false即关闭打印parpared语句的audit log。
set enable_prepared_stmt_audit_log=true;

package main

import (
	"database/sql"
	"fmt"
	"math/rand"
	"strings"
	"sync"
	"sync/atomic"
	"time"

	_ "github.com/go-sql-driver/mysql"
)

const (
	host     = "127.0.0.1"
	port     = 9038
	db       = "test"
	user     = "root"
	password = ""
	table    = "async_lineitem"
)

var (
	threadCount = 20
	batchSize   = 100
)

var totalInsertedRows int64
var rowsInsertedLastSecond int64

func main() {
	dbDSN := fmt.Sprintf("%s:%s@tcp(%s:%d)/%s?parseTime=true", user, password, host, port, db)
	db, err := sql.Open("mysql", dbDSN)
	if err != nil {
		fmt.Printf("Error opening database: %s\n", err)
		return
	}
	defer db.Close()

	var wg sync.WaitGroup
	for i := 0; i < threadCount; i++ {
		wg.Add(1)
		go func() {
			defer wg.Done()
			groupCommitInsertBatch(db)
		}()
	}

	go logInsertStatistics()

	wg.Wait()
}

func groupCommitInsertBatch(db *sql.DB) {
	for {
		valueStrings := make([]string, 0, batchSize)
		valueArgs := make([]interface{}, 0, batchSize*16)
		for i := 0; i < batchSize; i++ {
		    valueStrings = append(valueStrings, "(?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)")
			valueArgs = append(valueArgs, rand.Intn(1000))
			valueArgs = append(valueArgs, rand.Intn(1000))
			valueArgs = append(valueArgs, rand.Intn(1000))
			valueArgs = append(valueArgs, rand.Intn(1000))
			valueArgs = append(valueArgs, sql.NullFloat64{Float64: 1.0, Valid: true})
			valueArgs = append(valueArgs, sql.NullFloat64{Float64: 1.0, Valid: true})
			valueArgs = append(valueArgs, sql.NullFloat64{Float64: 1.0, Valid: true})
			valueArgs = append(valueArgs, sql.NullFloat64{Float64: 1.0, Valid: true})
			valueArgs = append(valueArgs, "N")
			valueArgs = append(valueArgs, "O")
			valueArgs = append(valueArgs, time.Now())
			valueArgs = append(valueArgs, time.Now())
			valueArgs = append(valueArgs, time.Now())
			valueArgs = append(valueArgs, "DELIVER IN PERSON")
			valueArgs = append(valueArgs, "SHIP")
			valueArgs = append(valueArgs, "N/A")
		}
		stmt := fmt.Sprintf("INSERT INTO %s VALUES %s",
			table, strings.Join(valueStrings, ","))
		_, err := db.Exec(stmt, valueArgs...)
		if err != nil {
			fmt.Printf("Error executing batch: %s\n", err)
			return
		}
		atomic.AddInt64(&rowsInsertedLastSecond, int64(batchSize))
		atomic.AddInt64(&totalInsertedRows, int64(batchSize))
	}
}

func logInsertStatistics() {
	for {
		time.Sleep(1 * time.Second)
		fmt.Printf("Total inserted rows: %d\n", totalInsertedRows)
		fmt.Printf("Rows inserted in the last second: %d\n", rowsInsertedLastSecond)
		rowsInsertedLastSecond = 0
	}
}

# 配置 session 变量开启 group commit (默认为 off_mode),开启异步模式
mysql> set group_commit = async_mode;

# 这里返回的 label 是 group_commit 开头的，可以区分出是否使用了 group commit
mysql> insert into dt values(1, 'Bob', 90), (2, 'Alice', 99);
Query OK, 2 rows affected (0.05 sec)
{'label':'group_commit_a145ce07f1c972fc-bd2c54597052a9ad', 'status':'PREPARE', 'txnId':'181508'}

# 可以看出这个 label, txn_id 和上一个相同，说明是攒到了同一个导入任务中
mysql> insert into dt(id, name) values(3, 'John');
Query OK, 1 row affected (0.01 sec)
{'label':'group_commit_a145ce07f1c972fc-bd2c54597052a9ad', 'status':'PREPARE', 'txnId':'181508'}

# 不能立刻查询到
mysql> select * from dt;
Empty set (0.01 sec)

# 10 秒后可以查询到，可以通过表属性 group_commit_interval 控制数据可见延迟。
mysql> select * from dt;
+------+-------+-------+
| id   | name  | score |
+------+-------+-------+
|    1 | Bob   |    90 |
|    2 | Alice |    99 |
|    3 | John  |  NULL |
+------+-------+-------+
3 rows in set (0.02 sec)

# 配置 session 变量开启 group commit (默认为 off_mode),开启同步模式
mysql> set group_commit = sync_mode;

# 这里返回的 label 是 group_commit 开头的，可以区分出是否谁用了 group commit，导入耗时至少是表属性 group_commit_interval。
mysql> insert into dt values(4, 'Bob', 90), (5, 'Alice', 99);
Query OK, 2 rows affected (10.06 sec)
{'label':'group_commit_d84ab96c09b60587_ec455a33cb0e9e87', 'status':'PREPARE', 'txnId':'3007', 'query_id':'fc6b94085d704a94-a69bfc9a202e66e2'}

# 数据可以立刻读出
mysql> select * from dt;
+------+-------+-------+
| id   | name  | score |
+------+-------+-------+
|    1 | Bob   |    90 |
|    2 | Alice |    99 |
|    3 | John  |  NULL |
|    4 | Bob   |    90 |
|    5 | Alice |    99 |
+------+-------+-------+
5 rows in set (0.03 sec)

mysql> set group_commit = off_mode;

6,Amy,60
7,Ross,98

# 导入时在 header 中增加"group_commit:async_mode"配置

curl --location-trusted -u {user}:{passwd} -T data.csv -H "group_commit:async_mode"  -H "column_separator:,"  http://{fe_host}:{http_port}/api/db/dt/_stream_load
{
    "TxnId": 7009,
    "Label": "group_commit_c84d2099208436ab_96e33fda01eddba8",
    "Comment": "",
    "GroupCommit": true,
    "Status": "Success",
    "Message": "OK",
    "NumberTotalRows": 2,
    "NumberLoadedRows": 2,
    "NumberFilteredRows": 0,
    "NumberUnselectedRows": 0,
    "LoadBytes": 19,
    "LoadTimeMs": 35,
    "StreamLoadPutTimeMs": 5,
    "ReadDataTimeMs": 0,
    "WriteDataTimeMs": 26
}

# 返回的 GroupCommit 为 true，说明进入了 group commit 的流程
# 返回的 Label 是 group_commit 开头的，是真正消费数据的导入关联的 label

# 导入时在 header 中增加"group_commit:sync_mode"配置

curl --location-trusted -u {user}:{passwd} -T data.csv -H "group_commit:sync_mode"  -H "column_separator:,"  http://{fe_host}:{http_port}/api/db/dt/_stream_load
{
    "TxnId": 3009,
    "Label": "group_commit_d941bf17f6efcc80_ccf4afdde9881293",
    "Comment": "",
    "GroupCommit": true,
    "Status": "Success",
    "Message": "OK",
    "NumberTotalRows": 2,
    "NumberLoadedRows": 2,
    "NumberFilteredRows": 0,
    "NumberUnselectedRows": 0,
    "LoadBytes": 19,
    "LoadTimeMs": 10044,
    "StreamLoadPutTimeMs": 4,
    "ReadDataTimeMs": 0,
    "WriteDataTimeMs": 10038
}

# 返回的 GroupCommit 为 true，说明进入了 group commit 的流程
# 返回的 Label 是 group_commit 开头的，是真正消费数据的导入关联的 label

# 修改提交间隔为 2 秒
ALTER TABLE dt SET ("group_commit_interval_ms" = "2000");

# 修改提交数据量为 128MB
ALTER TABLE dt SET ("group_commit_data_bytes" = "134217728");

CREATE TABLE IF NOT EXISTS lineitem (
  L_ORDERKEY    INTEGER NOT NULL,
  L_PARTKEY     INTEGER NOT NULL,
  L_SUPPKEY     INTEGER NOT NULL,
  L_LINENUMBER  INTEGER NOT NULL,
  L_QUANTITY    DECIMAL(15,2) NOT NULL,
  L_EXTENDEDPRICE  DECIMAL(15,2) NOT NULL,
  L_DISCOUNT    DECIMAL(15,2) NOT NULL,
  L_TAX         DECIMAL(15,2) NOT NULL,
  L_RETURNFLAG  CHAR(1) NOT NULL,
  L_LINESTATUS  CHAR(1) NOT NULL,
  L_SHIPDATE    DATE NOT NULL,
  L_COMMITDATE  DATE NOT NULL,
  L_RECEIPTDATE DATE NOT NULL,
  L_SHIPINSTRUCT CHAR(25) NOT NULL,
  L_SHIPMODE     CHAR(10) NOT NULL,
  L_COMMENT      VARCHAR(44) NOT NULL
)
DUPLICATE KEY(L_ORDERKEY, L_PARTKEY, L_SUPPKEY, L_LINENUMBER)
DISTRIBUTED BY HASH(L_ORDERKEY) BUCKETS 32
PROPERTIES (
  "replication_num" = "3"
);

CREATE TABLE IF NOT EXISTS lineitem (
  L_ORDERKEY    INTEGER NOT NULL,
  L_PARTKEY     INTEGER NOT NULL,
  L_SUPPKEY     INTEGER NOT NULL,
  L_LINENUMBER  INTEGER NOT NULL,
  L_QUANTITY    DECIMAL(15,2) NOT NULL,
  L_EXTENDEDPRICE  DECIMAL(15,2) NOT NULL,
  L_DISCOUNT    DECIMAL(15,2) NOT NULL,
  L_TAX         DECIMAL(15,2) NOT NULL,
  L_RETURNFLAG  CHAR(1) NOT NULL,
  L_LINESTATUS  CHAR(1) NOT NULL,
  L_SHIPDATE    DATE NOT NULL,
  L_COMMITDATE  DATE NOT NULL,
  L_RECEIPTDATE DATE NOT NULL,
  L_SHIPINSTRUCT CHAR(25) NOT NULL,
  L_SHIPMODE     CHAR(10) NOT NULL,
  L_COMMENT      VARCHAR(44) NOT NULL
)
DUPLICATE KEY(L_ORDERKEY, L_PARTKEY, L_SUPPKEY, L_LINENUMBER)
DISTRIBUTED BY HASH(L_ORDERKEY) BUCKETS 32
PROPERTIES (
  "replication_num" = "3"
);