Spring事務隔離級別和傳播特性

相信每一個人都被問過無數次Spring聲明式事務的隔離級別和傳播機制吧！今天我也來講說這兩個東西.

加入一個小插曲，
一天電話裏有人問我聲明式事務隔離級別有哪幾種，
我就回答了7種，
他問我Spring的版本，
我回答爲3.0。
他說那應該是2.5的，3.0好像變少了。
我回答這個沒有確認過。

後來我就google了一下，沒發現什麼痕跡說明事務的隔離級別變少了，也查了下官方文檔，也沒有相關的說明。索性在github上clone一下Spring的源碼，看看源碼中有幾種就是幾種了唄。

後來想一想那天他那麼問我徹底多是一個坑啊，由於交談的過程當中挖過至少兩個坑了。再者說，Spring要向下兼容的，若是少了怎麼處理呢？固然這兩點都是我本身的猜想。

聲明式事務

在Spring中，聲明式事務是用事務參數來定義的。一個事務參數就是對事務策略應該如何應用到某個方法的一段描述，以下圖所示一個事務參數共有5個方面組成：

傳播行爲

事務的第一個方面是傳播行爲。傳播行爲定義關於客戶端和被調用方法的事務邊界。Spring定義了7中傳播行爲。

傳播行爲	意義
PROPAGATION_MANDATORY	表示該方法必須運行在一個事務中。若是當前沒有事務正在發生，將拋出一個異常
PROPAGATION_NESTED	表示若是當前正有一個事務在進行中，則該方法應當運行在一個嵌套式事務中。被嵌套的事務能夠獨立於封裝事務進行提交或回滾。若是封裝事務不存在，行爲就像PROPAGATION_REQUIRES同樣。
PROPAGATION_NEVER	表示當前的方法不該該在一個事務中運行。若是一個事務正在進行，則會拋出一個異常。
PROPAGATION_NOT_SUPPORTED	表示該方法不該該在一個事務中運行。若是一個現有事務正在進行中，它將在該方法的運行期間被掛起。
PROPAGATION_SUPPORTS	表示當前方法不須要事務性上下文，可是若是有一個事務已經在運行的話，它也能夠在這個事務裏運行。
PROPAGATION_REQUIRES_NEW	表示當前方法必須在它本身的事務裏運行。一個新的事務將被啓動，並且若是有一個現有事務在運行的話，則將在這個方法運行期間被掛起。
PROPAGATION_REQUIRES	表示當前方法必須在一個事務中運行。若是一個現有事務正在進行中，該方法將在那個事務中運行，不然就要開始一個新事務。

傳播規則回答了這樣一個問題，就是一個新的事務應該被啓動仍是被掛起，或者是一個方法是否應該在事務性上下文中運行。

隔離級別

聲明式事務的第二個方面是隔離級別。隔離級別定義一個事務可能受其餘併發事務活動活動影響的程度。另外一種考慮一個事務的隔離級別的方式，是把它想象爲那個事務對於事物處理數據的自私程度。

在一個典型的應用程序中，多個事務同時運行，常常會爲了完成他們的工做而操做同一個數據。併發雖然是必需的，可是會致使一下問題：

• 髒讀（Dirty read）-- 髒讀發生在一個事務讀取了被另外一個事務改寫但還沒有提交的數據時。若是這些改變在稍後被回滾了，那麼第一個事務讀取的數據就會是無效的。
• 不可重複讀（Nonrepeatable read）-- 不可重複讀發生在一個事務執行相同的查詢兩次或兩次以上，但每次查詢結果都不相同時。這一般是因爲另外一個併發事務在兩次查詢之間更新了數據。
• 幻影讀（Phantom reads）-- 幻影讀和不可重複讀類似。當一個事務（T1）讀取幾行記錄後，另外一個併發事務（T2）插入了一些記錄時，幻影讀就發生了。在後來的查詢中，第一個事務（T1）就會發現一些原來沒有的額外記錄。

在理想狀態下，事務之間將徹底隔離，從而能夠防止這些問題發生。然而，徹底隔離會影響性能，由於隔離常常牽扯到鎖定在數據庫中的記錄（並且有時是鎖定完整的數據表）。侵佔性的鎖定會阻礙併發，要求事務相互等待來完成工做。

考慮到徹底隔離會影響性能，並且並非全部應用程序都要求徹底隔離，因此有時能夠在事務隔離方面靈活處理。所以，就會有好幾個隔離級別。

隔離級別	含義
ISOLATION_DEFAULT	使用後端數據庫默認的隔離級別。
ISOLATION_READ_UNCOMMITTED	容許讀取還沒有提交的更改。可能致使髒讀、幻影讀或不可重複讀。
ISOLATION_READ_COMMITTED	容許從已經提交的併發事務讀取。可防止髒讀，但幻影讀和不可重複讀仍可能會發生。
ISOLATION_REPEATABLE_READ	對相同字段的屢次讀取的結果是一致的，除非數據被當前事務自己改變。可防止髒讀和不可重複讀，但幻影讀仍可能發生。
ISOLATION_SERIALIZABLE	徹底服從ACID的隔離級別，確保不發生髒讀、不可重複讀和幻影讀。這在全部隔離級別中也是最慢的，由於它一般是經過徹底鎖定當前事務所涉及的數據表來完成的。

只讀

聲明式事務的第三個特性是它是不是一個只讀事務。若是一個事務只對後端數據庫執行讀操做，那麼該數據庫就可能利用那個事務的只讀特性，採起某些優化 措施。經過把一個事務聲明爲只讀，能夠給後端數據庫一個機會來應用那些它認爲合適的優化措施。因爲只讀的優化措施是在一個事務啓動時由後端數據庫實施的， 所以，只有對於那些具備可能啓動一個新事務的傳播行爲（PROPAGATION_REQUIRES_NEW、PROPAGATION_REQUIRED、 ROPAGATION_NESTED）的方法來講，將事務聲明爲只讀纔有意義。

此外，若是使用Hibernate做爲持久化機制，那麼把一個事務聲明爲只讀，將使Hibernate的flush模式被設置爲FLUSH_NEVER。這就告訴Hibernate避免和數據庫進行沒必要要的對象同步，從而把全部更新延遲到事務的結束。

事務超時

爲了使一個應用程序很好地執行，它的事務不能運行太長時間。所以，聲明式事務的下一個特性就是它的超時。

假設事務的運行時間變得格外的長，因爲事務可能涉及對後端數據庫的鎖定，因此長時間運行的事務會沒必要要地佔用數據庫資源。這時就能夠聲明一個事務在特定秒數後自動回滾，沒必要等它本身結束。

因爲超時時鐘在一個事務啓動的時候開始的，所以，只有對於那些具備可能啓動一個新事務的傳播行爲（PROPAGATION_REQUIRES_NEW、PROPAGATION_REQUIRED、ROPAGATION_NESTED）的方法來講，聲明事務超時纔有意義。

回滾規則

事務五邊形的對後一個邊是一組規則，它們定義哪些異常引發回滾，哪些不引發。在默認設置下，事務只在出現運行時異常（runtime exception）時回滾，而在出現受檢查異常（checked exception）時不回滾（這一行爲和EJB中的回滾行爲是一致的）。

不過，也能夠聲明在出現特定受檢查異常時像運行時異常同樣回滾。一樣，也能夠聲明一個事務在出現特定的異常時不回滾，即便那些異常是運行時一場。

擴展閱讀

標題是隻有事務的隔離級別和傳播機制，卻順帶這把聲明式事務的五個特性都講述了一遍。：）

文章開頭說過查看Spring中事務的源碼來確認3.0版本及以後事務的傳播機制是否減小了，其實在TransactionDefinition這個接口中定義了事務的隔離級別、傳播機制、只讀以及超時相關的所有信息。源碼以下，感興趣的能夠本身對照一下，看看英文註釋。

/*
 * Copyright 2002-2010 the original author or authors.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package org.springframework.transaction;

import java.sql.Connection;

/**
 * Interface that defines Spring-compliant transaction properties.
 * Based on the propagation behavior definitions analogous to EJB CMT attributes.
 *
 * <p>Note that isolation level and timeout settings will not get applied unless
 * an actual new transaction gets started. As only {@link #PROPAGATION_REQUIRED},
 * {@link #PROPAGATION_REQUIRES_NEW} and {@link #PROPAGATION_NESTED} can cause
 * that, it usually doesn't make sense to specify those settings in other cases.
 * Furthermore, be aware that not all transaction managers will support those
 * advanced features and thus might throw corresponding exceptions when given
 * non-default values.
 *
 * <p>The {@link #isReadOnly() read-only flag} applies to any transaction context,
 * whether backed by an actual resource transaction or operating non-transactionally
 * at the resource level. In the latter case, the flag will only apply to managed
 * resources within the application, such as a Hibernate <code>Session</code>.
 *
 * @author Juergen Hoeller
 * @since 08.05.2003
 * @see PlatformTransactionManager#getTransaction(TransactionDefinition)
 * @see org.springframework.transaction.support.DefaultTransactionDefinition
 * @see org.springframework.transaction.interceptor.TransactionAttribute
 */
public interface TransactionDefinition {

  /**
   * Support a current transaction; create a new one if none exists.
   * Analogous to the EJB transaction attribute of the same name.
   * <p>This is typically the default setting of a transaction definition,
   * and typically defines a transaction synchronization scope.
   */
  int PROPAGATION_REQUIRED = 0;

  /**
   * Support a current transaction; execute non-transactionally if none exists.
   * Analogous to the EJB transaction attribute of the same name.
   * <p><b>NOTE:</b> For transaction managers with transaction synchronization,
   * <code>PROPAGATION_SUPPORTS</code> is slightly different from no transaction
   * at all, as it defines a transaction scope that synchronization might apply to.
   * As a consequence, the same resources (a JDBC <code>Connection</code>, a
   * Hibernate <code>Session</code>, etc) will be shared for the entire specified
   * scope. Note that the exact behavior depends on the actual synchronization
   * configuration of the transaction manager!
   * <p>In general, use <code>PROPAGATION_SUPPORTS</code> with care! In particular, do
   * not rely on <code>PROPAGATION_REQUIRED</code> or <code>PROPAGATION_REQUIRES_NEW</code>
   * <i>within</i> a <code>PROPAGATION_SUPPORTS</code> scope (which may lead to
   * synchronization conflicts at runtime). If such nesting is unavoidable, make sure
   * to configure your transaction manager appropriately (typically switching to
   * "synchronization on actual transaction").
   * @see org.springframework.transaction.support.AbstractPlatformTransactionManager#setTransactionSynchronization
   * @see org.springframework.transaction.support.AbstractPlatformTransactionManager#SYNCHRONIZATION_ON_ACTUAL_TRANSACTION
   */
  int PROPAGATION_SUPPORTS = 1;

  /**
   * Support a current transaction; throw an exception if no current transaction
   * exists. Analogous to the EJB transaction attribute of the same name.
   * <p>Note that transaction synchronization within a <code>PROPAGATION_MANDATORY</code>
   * scope will always be driven by the surrounding transaction.
   */
  int PROPAGATION_MANDATORY = 2;

  /**
   * Create a new transaction, suspending the current transaction if one exists.
   * Analogous to the EJB transaction attribute of the same name.
   * <p><b>NOTE:</b> Actual transaction suspension will not work out-of-the-box
   * on all transaction managers. This in particular applies to
   * {@link org.springframework.transaction.jta.JtaTransactionManager},
   * which requires the <code>javax.transaction.TransactionManager</code>
   * to be made available it to it (which is server-specific in standard J2EE).
   * <p>A <code>PROPAGATION_REQUIRES_NEW</code> scope always defines its own
   * transaction synchronizations. Existing synchronizations will be suspended
   * and resumed appropriately.
   * @see org.springframework.transaction.jta.JtaTransactionManager#setTransactionManager
   */
  int PROPAGATION_REQUIRES_NEW = 3;

  /**
   * Do not support a current transaction; rather always execute non-transactionally.
   * Analogous to the EJB transaction attribute of the same name.
   * <p><b>NOTE:</b> Actual transaction suspension will not work out-of-the-box
   * on all transaction managers. This in particular applies to
   * {@link org.springframework.transaction.jta.JtaTransactionManager},
   * which requires the <code>javax.transaction.TransactionManager</code>
   * to be made available it to it (which is server-specific in standard J2EE).
   * <p>Note that transaction synchronization is <i>not</i> available within a
   * <code>PROPAGATION_NOT_SUPPORTED</code> scope. Existing synchronizations
   * will be suspended and resumed appropriately.
   * @see org.springframework.transaction.jta.JtaTransactionManager#setTransactionManager
   */
  int PROPAGATION_NOT_SUPPORTED = 4;

  /**
   * Do not support a current transaction; throw an exception if a current transaction
   * exists. Analogous to the EJB transaction attribute of the same name.
   * <p>Note that transaction synchronization is <i>not</i> available within a
   * <code>PROPAGATION_NEVER</code> scope.
   */
  int PROPAGATION_NEVER = 5;

  /**
   * Execute within a nested transaction if a current transaction exists,
   * behave like {@link #PROPAGATION_REQUIRED} else. There is no analogous
   * feature in EJB.
   * <p><b>NOTE:</b> Actual creation of a nested transaction will only work on
   * specific transaction managers. Out of the box, this only applies to the JDBC
   * {@link org.springframework.jdbc.datasource.DataSourceTransactionManager}
   * when working on a JDBC 3.0 driver. Some JTA providers might support
   * nested transactions as well.
   * @see org.springframework.jdbc.datasource.DataSourceTransactionManager
   */
  int PROPAGATION_NESTED = 6;


  /**
   * Use the default isolation level of the underlying datastore.
   * All other levels correspond to the JDBC isolation levels.
   * @see java.sql.Connection
   */
  int ISOLATION_DEFAULT = -1;

  /**
   * Indicates that dirty reads, non-repeatable reads and phantom reads
   * can occur.
   * <p>This level allows a row changed by one transaction to be read by another
   * transaction before any changes in that row have been committed (a "dirty read").
   * If any of the changes are rolled back, the second transaction will have
   * retrieved an invalid row.
   * @see java.sql.Connection#TRANSACTION_READ_UNCOMMITTED
   */
  int ISOLATION_READ_UNCOMMITTED = Connection.TRANSACTION_READ_UNCOMMITTED;

  /**
   * Indicates that dirty reads are prevented; non-repeatable reads and
   * phantom reads can occur.
   * <p>This level only prohibits a transaction from reading a row
   * with uncommitted changes in it.
   * @see java.sql.Connection#TRANSACTION_READ_COMMITTED
   */
  int ISOLATION_READ_COMMITTED = Connection.TRANSACTION_READ_COMMITTED;

  /**
   * Indicates that dirty reads and non-repeatable reads are prevented;
   * phantom reads can occur.
   * <p>This level prohibits a transaction from reading a row with uncommitted changes
   * in it, and it also prohibits the situation where one transaction reads a row,
   * a second transaction alters the row, and the first transaction re-reads the row,
   * getting different values the second time (a "non-repeatable read").
   * @see java.sql.Connection#TRANSACTION_REPEATABLE_READ
   */
  int ISOLATION_REPEATABLE_READ = Connection.TRANSACTION_REPEATABLE_READ;

  /**
   * Indicates that dirty reads, non-repeatable reads and phantom reads
   * are prevented.
   * <p>This level includes the prohibitions in {@link #ISOLATION_REPEATABLE_READ}
   * and further prohibits the situation where one transaction reads all rows that
   * satisfy a <code>WHERE</code> condition, a second transaction inserts a row
   * that satisfies that <code>WHERE</code> condition, and the first transaction
   * re-reads for the same condition, retrieving the additional "phantom" row
   * in the second read.
   * @see java.sql.Connection#TRANSACTION_SERIALIZABLE
   */
  int ISOLATION_SERIALIZABLE = Connection.TRANSACTION_SERIALIZABLE;


  /**
   * Use the default timeout of the underlying transaction system,
   * or none if timeouts are not supported. 
   */
  int TIMEOUT_DEFAULT = -1;


  /**
   * Return the propagation behavior.
   * <p>Must return one of the <code>PROPAGATION_XXX</code> constants
   * defined on {@link TransactionDefinition this interface}.
   * @return the propagation behavior
   * @see #PROPAGATION_REQUIRED
   * @see org.springframework.transaction.support.TransactionSynchronizationManager#isActualTransactionActive()
   */
  int getPropagationBehavior();

  /**
   * Return the isolation level.
   * <p>Must return one of the <code>ISOLATION_XXX</code> constants
   * defined on {@link TransactionDefinition this interface}.
   * <p>Only makes sense in combination with {@link #PROPAGATION_REQUIRED}
   * or {@link #PROPAGATION_REQUIRES_NEW}.
   * <p>Note that a transaction manager that does not support custom isolation levels
   * will throw an exception when given any other level than {@link #ISOLATION_DEFAULT}.
   * @return the isolation level
   */
  int getIsolationLevel();

  /**
   * Return the transaction timeout.
   * <p>Must return a number of seconds, or {@link #TIMEOUT_DEFAULT}.
   * <p>Only makes sense in combination with {@link #PROPAGATION_REQUIRED}
   * or {@link #PROPAGATION_REQUIRES_NEW}.
   * <p>Note that a transaction manager that does not support timeouts will throw
   * an exception when given any other timeout than {@link #TIMEOUT_DEFAULT}.
   * @return the transaction timeout
   */
  int getTimeout();

  /**
   * Return whether to optimize as a read-only transaction.
   * <p>The read-only flag applies to any transaction context, whether
   * backed by an actual resource transaction
   * ({@link #PROPAGATION_REQUIRED}/{@link #PROPAGATION_REQUIRES_NEW}) or
   * operating non-transactionally at the resource level
   * ({@link #PROPAGATION_SUPPORTS}). In the latter case, the flag will
   * only apply to managed resources within the application, such as a
   * Hibernate <code>Session</code>.
   * <p>This just serves as a hint for the actual transaction subsystem;
   * it will <i>not necessarily</i> cause failure of write access attempts.
   * A transaction manager which cannot interpret the read-only hint will
   * <i>not</i> throw an exception when asked for a read-only transaction.
   * @return <code>true</code> if the transaction is to be optimized as read-only 
   * @see org.springframework.transaction.support.TransactionSynchronization#beforeCommit(boolean)
   * @see org.springframework.transaction.support.TransactionSynchronizationManager#isCurrentTransactionReadOnly()
   */
  boolean isReadOnly();

  /**
   * Return the name of this transaction. Can be <code>null</code>.
   * <p>This will be used as the transaction name to be shown in a
   * transaction monitor, if applicable (for example, WebLogic's).
   * <p>In case of Spring's declarative transactions, the exposed name will be
   * the <code>fully-qualified class name + "." + method name</code> (by default).
   * @return the name of this transaction
   * @see org.springframework.transaction.interceptor.TransactionAspectSupport
   * @see org.springframework.transaction.support.TransactionSynchronizationManager#getCurrentTransactionName()
   */
  String getName();

}

 

仍是以爲不安心，發兩張圖證實隔離級別和傳播機制：

• eclipse中給出的關於傳播機制的智能提示截圖

• eclipse中給出的關於隔離級別的智能提示截圖