最近的项目中发现某些页面打开速度非常慢,最慢的情况下甚至到了30秒。
分析原因后发现是加载由于实体间配了1对1的双向关联,导致lazy加载失效,每次加载都会把关联实体的全部属性查询出来,对性能影响特别大。
分析原因后发现是加载由于实体间配了1对1的双向关联,导致lazy加载失效,每次加载都会把关联实体的全部属性查询出来,对性能影响特别大。
最终的解决方案是直接执行sql取得需要的字段,然后再转为对象。代码如下:
Session session = null;
String sql = "select o.ID as id, o.OFFICE_NO as officeNo from ASY_OFFICE_INFO o where o.OFFICE_SYSTEM='" + sys + "' and o.OFFICE_STATUS = '1' and o.DEPARTMENT_ID =" + company + " and o.ISEXISTS is null";
session = baseDao.getHibernate().getSessionFactory().openSession();
SQLQuery query = session.createSQLQuery(sql);
query.addScalar("id", Hibernate.LONG).addScalar(
"officeNo", Hibernate.STRING);
query.setResultTransformer(Transformers
.aliasToBean(ManagerOfficeInfo.class));
List<ManagerOfficeInfo> list = query.list();
session.close();===========另一种实现方式的分割线=========
private static DataSource dataSource = null;
/**
* 获取系列生成器数据库连接,无论哪种用户都取同一个连接,确保hibernate缓存正确;.
*
* @return the connection
* @throws SQLException the sQL exception
*/
private static Connection getConnection() throws SQLException {
if (dataSource == null) {
dataSource = (DataSource) ApplicationContextKeeper.getAppCtx().getBean("dataSource");
}
return dataSource.getConnection();
}
/**
* Gather over bill.
*
* @param dataList the data list
* @param flowMonth the flow month
* @throws Exception the exception
*/
private void gatherOverBill(List<IbeFee> dataList, String flowMonth) throws Exception {
// PreparedStatement pstm = null;
PreparedStatement pstm2 = null;
// ResultSet rs = null;
Connection cn = null;
int size = dataList.size();
try {
cn = getConnection();
/*
* StringBuffer sb = new StringBuffer(); sb.append(" select t.branch , t.agent , t.bill_month , t.total_charge , t.id from asy_ibe_fee t
* where t.matching = '1'"); sb.append(" and t.bill_month = '" + flowMonth +"'");
*/
StringBuffer sb2 = new StringBuffer();
sb2.append(" insert into asy_ibe_bill (ID,BRANCH,AGENT,BILL_MONTH,BILL_AMOUNT,AUDITING,");
sb2.append(" REFUNDED,CALCULATOR_NO,REMARKS,MAIN_ID)");
sb2.append(" values (asy_ibe_fee_id.nextval ,? ,? ,? ,? ,'0' ,'0' ,'' ,'' , ? )");
pstm2 = cn.prepareStatement(sb2.toString());
// pstm = cn.prepareStatement(sb.toString());
// rs = pstm.executeQuery();
for (int i = 0; i < size; i++) {
IbeFee ibe = dataList.get(i);
if (!"1".equals(ibe.getMATCHING())) {
continue;
}
pstm2.setString(1, ibe.getBranch().getId().toString());
pstm2.setString(2, ibe.getDepartment().getId().toString());
pstm2.setString(3, ibe.getBILL_MONTH());
pstm2.setString(4, ibe.getTOTAL_CHARGE());
pstm2.setString(5, ibe.getId().toString());
pstm2.addBatch();
if (i > 0 && i % 1000 == 0) {
pstm2.executeBatch();
pstm2.clearBatch();
}
}
/*
* while(rs.next()){ pstm2.setString(1, rs.getString("branch")); pstm2.setString(2, rs.getString("agent")); pstm2.setString(3,
* rs.getString("bill_month")); pstm2.setString(4, rs.getString("total_charge")); pstm2.setString(5, rs.getString("id"));
* pstm2.addBatch();
*
* i++; if(i>0 && i%1000==0){ pstm2.executeBatch(); pstm2.clearBatch(); } }
*/
pstm2.executeBatch();
// cn.commit();
} catch (Exception ex) {
ex.printStackTrace();
throw new Exception(ex);
} finally {
if (pstm2 != null) {
pstm2.close();
}
// if(pstm!=null) pstm.close();
if (cn != null) {
cn.close();
}
}
}
/**
* Delete.
*
* @param flowMonth the flow month
* @throws Exception the exception
*/
private void delete(String flowMonth) throws Exception {
PreparedStatement pstm = null;
Connection cn = null;
try {
cn = getConnection();
String sql = "delete from ASY_IBE_FEE where BILL_MONTH=?";
pstm = cn.prepareStatement(sql);
pstm.setString(1, flowMonth);
pstm.execute();
pstm.close();
sql = "delete from ASY_IBE_BILL where BILL_MONTH=?";
pstm = cn.prepareStatement(sql);
pstm.setString(1, flowMonth);
pstm.execute();
// cn.commit();
} catch (Exception ex) {
ex.printStackTrace();
throw new Exception(ex);
} finally {
if (cn != null) {
cn.close();
}
if (pstm != null) {
pstm.close();
}
}
}
/**
* 取得C系统的所有office.
*
* @return the map
* @throws Exception the exception
*/
protected Map<String, OfficeInfo> loadOfficeInfos() throws Exception {
Map<String, OfficeInfo> map = new HashMap<String, OfficeInfo>();
Connection cn = null;
PreparedStatement pstm = null;
ResultSet rs = null;
try {
StringBuffer sb = new StringBuffer();
sb.append(" select id,office_status,office_system,office_no,department_id,branch_id ");
sb.append(" from ASY_OFFICE_INFO where office_status='1' and office_system='1' and (isexists <> '1' or isexists is null) ");
cn = getConnection();
pstm = cn.prepareStatement(sb.toString());
rs = pstm.executeQuery();
while (rs.next()) {
String officeNo = StringUtils.trimToEmpty(rs.getString("office_no")).toUpperCase();
OfficeInfo o = new OfficeInfo();
o.setBranchId(rs.getLong("branch_id"));
o.setDepartmentId(rs.getLong("department_id"));
o.setOfficeNo(officeNo);
o.setOfficeSystem(StringUtils.trimToEmpty(rs.getString("office_system")));
if (StringUtils.isNotEmpty(officeNo) && !map.containsKey(officeNo)) {
map.put(officeNo, o);
}
}
} catch (Exception ex) {
ex.printStackTrace();
throw new Exception(ex);
} finally {
if (rs != null) {
rs.close();
}
if (pstm != null) {
pstm.close();
}
if (cn != null) {
cn.close();
}
}
return map;
}
针对Hibernate性能问题,采用直接执行SQL获取特定字段再转换为对象的策略,以此提高查询效率。
3295

被折叠的 条评论
为什么被折叠?



