MySQL JDBC FetchSize解析

根据http://boylook.blog.51cto.com/7934327/1298634提到MySQL JDBC的fetchsize问题. 在MySQl官方文档里只提到了streaming模式和fetchALL两种模式，那么是不是就没有中间的状态呢？

首先是看Java JDBC的API查看setFetchSize:

setFetchSize(int rows) 
Givesthe JDBC driver a hint as to the number of rows that should be fetched from thedatabase when more rows are needed for this ResultSetobject.

查看的结果给出这里的rows只是一个hint，那么对于MySQL JDBC来说是如何实现的呢？

查到MySQL一个“bug”http://bugs.mysql.com/bug.php?id=18148里说：

There is experimental support for fetching rows in batches

When using Connector/J 5.0.1 along with more recent builds of the MySQL server, you can add "useCursorFetch=true" to your JDBC url parameters, and the driver will fetch rows in batches of size setFetchSize() as defined in the JDBC API.

One could also argue that the behavior _does_ follow the JDBC API, quoting from the APIDOCS for Statement.setFetchSize():

"Gives the JDBC driver a hint as to the number of rows that should be fetched from the database when more rows are needed. The number of rows specified affects only result sets created using this statement. If the value specified is zero, then the hint is ignored. The default value is zero."

It's only a _hint_. Driver vendors are allowed to ignore hints. The very reason that the wording is there is because there are quite a few vendors who can not adhere to this "contract" in all situations.

又说是experiment，又说是hint可能被ignore，到底支持还不不支持呢？并且官方文档给出的这个参数没说到底是不是会忽略掉fetchSize.按照故事的尿性来说我该看MySQL JDBC的源码了:

1.首先判断是否可以进行cursor read

if (this.connection.versionMeetsMinimum(5, 0, 2)

 && this.connection.getUseCursorFetch()

 && isBinaryEncoded

 && callingStatement != null

 && callingStatement.getFetchSize() != 0

 && callingStatement.getResultSetType() == ResultSet.TYPE_FORWARD_ONLY)

若OK，则RowData rows = new RowDataCursor

2.如果不满足，则判断是否可以进行Streaming Read：

If(this.resultSetType== java.sql.ResultSet.TYPE_FORWARD_ONLY)

   && (this.resultSetConcurrency == java.sql.ResultSet.CONCUR_READ_ONLY) && (this.fetchSize == Integer.MIN_VALUE)

如果不满足则rowData = readSingleRowSet

3.否则rowData = new RowDataDynamic

这里主要关注RowDataCursor的next实现方式：next会去调用fetchMoreRows，

3.1 如果fetchsize==Integer.MIN_VALUE,则另fetchsize=1，类似stream 模式

3.2 否则回调fetchRowsViaCursor：

this.sharedSendPacket.writeByte((byte) MysqlDefs.COM_FETCH);

 this.sharedSendPacket.writeLong(statementId);

 this.sharedSendPacket.writeLong(fetchSize);

 sendCommand(MysqlDefs.COM_FETCH, null, this.sharedSendPacket, true,

 null);

 while ((row = nextRow(columnTypes, columnTypes.length, true,

 ResultSet.CONCUR_READ_ONLY, false, useBufferRowExplicit, false, null)) != null) {

 fetchedRows.add(row);

}

当执行sendCommand后，看MySQL源码是如何处理的：

317 void Materialized_cursor::fetch(ulong num_rows)

 318 {

 319 THD *thd= table->in_use;

 320 

 321 int res=0;

 322 result->begin_dataset();

 323 for(fetch_limit+= num_rows; fetch_count < fetch_limit; fetch_count++)

 324 {

 325 if((res= table->file->rnd_next(table->record[0])))

 326 break;

 327 /* Send data only if the read was successful. */

 328 /*

 329 If network write failed (i.e. due to a closed socked),

 330 the error has already been set. Just return.

 331 */

 332 if(result->send_data(item_list))

 333 return;

 334 }

 335 

 336 switch(res) {

 337 case0:

 338 thd->server_status|= SERVER_STATUS_CURSOR_EXISTS;

 339 result->send_eof();

 340 break;

 341 case HA_ERR_END_OF_FILE:

 342 thd->server_status|= SERVER_STATUS_LAST_ROW_SENT;

 343 result->send_eof();

 344 close();

 345 break;

 346 default:

 347 table->file->print_error(res,MYF(0));

 348 close();

 349 break;

 350 }

 351 }

分析到这里基本上可以确定MySQL也是支持batch fetch的.

本文转自MIKE老毕 51CTO博客，原文链接：http://blog.51cto.com/boylook/1308511，如需转载请自行联系原作者