Tablestore：離線寫入資料

前提條件

安裝 Tablestore Java SDK並初始化用戶端。

功能說明

public BulkImportResponse bulkImport(BulkImportRequest bulkImportRequest) throws TableStoreException, ClientException

在單個請求中提交多行資料操作，由服務端按行獨立寫入並按行返回結果，適用於巨量資料情境下的離線大量匯入。

• 通過 addRowChange(RowChange) 或 addRowChanges(List<RowChange>) 添加行操作；支援 RowPutChange（插入）、RowUpdateChange（修改）、RowDeleteChange（刪除）三種子類，可在同一請求中混合使用。

• 通過 BulkImportResponse.getResult(succeedRows, failedRows) 拿成功和失敗的行；isAllSucceed() 判斷全量成功。每行結果獨立，部分行失敗不影響其他行寫入。

• BulkImportRequest.createRequestForRetry(failedRows) 基於失敗行產生只包含失敗行的重試請求，用於按行層級重試。

以下樣本將 5 行資料批量插入資料表 bulk_import_demo，並列印成功和失敗行數。

String tableName = "bulk_import_demo";

BulkImportRequest request = new BulkImportRequest(tableName);

List<RowChange> rowChanges = new ArrayList<RowChange>();
for (int i = 0; i < 5; i++) {
    PrimaryKey pk = PrimaryKeyBuilder.createPrimaryKeyBuilder()
            .addPrimaryKeyColumn("pk", PrimaryKeyValue.fromString("row" + i))
            .build();
    RowPutChange put = new RowPutChange(tableName, pk);
    put.addColumn(new Column("col1", ColumnValue.fromString("v" + i)));
    rowChanges.add(put);
}
request.addRowChanges(rowChanges);

BulkImportResponse response = client.bulkImport(request);

// 通過 succeedRows / failedRows 拿成功和失敗的行
List<BulkImportResponse.RowResult> succeedRows = new ArrayList<BulkImportResponse.RowResult>();
List<BulkImportResponse.RowResult> failedRows = new ArrayList<BulkImportResponse.RowResult>();
response.getResult(succeedRows, failedRows);

System.out.println("All succeed: " + response.isAllSucceed());
System.out.println("Succeed: " + succeedRows.size() + ", Failed: " + failedRows.size());

參數說明

情境樣本

String tableName = "bulk_import_demo";

BulkImportRequest request = new BulkImportRequest(tableName);

// 插入新行
PrimaryKey pkPut = PrimaryKeyBuilder.createPrimaryKeyBuilder()
        .addPrimaryKeyColumn("pk", PrimaryKeyValue.fromString("mixed_put"))
        .build();
RowPutChange put = new RowPutChange(tableName, pkPut);
put.addColumn(new Column("col1", ColumnValue.fromString("put_value")));
request.addRowChange(put);

// 更新已有行（追加列）
PrimaryKey pkUpdate = PrimaryKeyBuilder.createPrimaryKeyBuilder()
        .addPrimaryKeyColumn("pk", PrimaryKeyValue.fromString("row0"))
        .build();
RowUpdateChange update = new RowUpdateChange(tableName, pkUpdate);
update.put(new Column("col2", ColumnValue.fromLong(100)));
request.addRowChange(update);

// 刪除已有行
PrimaryKey pkDelete = PrimaryKeyBuilder.createPrimaryKeyBuilder()
        .addPrimaryKeyColumn("pk", PrimaryKeyValue.fromString("row1"))
        .build();
RowDeleteChange delete = new RowDeleteChange(tableName, pkDelete);
request.addRowChange(delete);

BulkImportResponse response = client.bulkImport(request);
System.out.println("Mixed all succeed: " + response.isAllSucceed());

String tableName = "bulk_import_demo";

BulkImportRequest request = new BulkImportRequest(tableName);

PrimaryKey pk = PrimaryKeyBuilder.createPrimaryKeyBuilder()
        .addPrimaryKeyColumn("pk", PrimaryKeyValue.fromString("good"))
        .build();
RowPutChange row = new RowPutChange(tableName, pk);
row.addColumn(new Column("col1", ColumnValue.fromString("ok")));
request.addRowChange(row);

BulkImportResponse response = client.bulkImport(request);

List<BulkImportResponse.RowResult> succeedRows = new ArrayList<BulkImportResponse.RowResult>();
List<BulkImportResponse.RowResult> failedRows = new ArrayList<BulkImportResponse.RowResult>();
response.getResult(succeedRows, failedRows);

// 僅當存在失敗行時，基於失敗行產生只包含失敗行的重試請求
if (!failedRows.isEmpty()) {
    BulkImportRequest retryRequest = request.createRequestForRetry(failedRows);
    BulkImportResponse retryResponse = client.bulkImport(retryRequest);
    System.out.println("Retry all succeed: " + retryResponse.isAllSucceed());
}

相關文檔

• 批量資料操作

• 離線讀取資料