NTP方式保证以时间戳同步可靠性

数据库同步的方式有设置标志位同步方式、以时间戳同步的方式，对于一对一的同步这两种方式都满足，但是对于一对多的同步，则只能选择时间戳的同步方式了，但是已时间戳同步的方式的一个问题是如何可靠的保证数据可以不丢失的同步到数据库中。下面有两种方式

数据库同步的方式有设置标志位同步方式、以时间戳同步的方式，对于一对一的同步这两种方式都满足，但是对于一对多的同步，则只能选择时间戳的同步方式了，但是已时间戳同步的方式的一个问题是如何可靠的保证数据可以不丢失的同步到数据库中。下面有两种方式来保证以时间戳同步的可靠性。

1、计算服务器与客户端两台电脑的时间差，将该时间差记入在同步时间的比对当中，下图是使用kettle做的一个计算同步的时间差的思路：

该思路的xml代码为：

<transformation>
  <info>
    <name>getLeadValue</name>
    <description/>
    <extended_description/>
    <trans_version/>
    <trans_type>Normal</trans_type>
    <trans_status>0</trans_status>
    <directory>/downloadServerData/commonData</directory>
    <parameters>
    </parameters>
    <log>
<trans-log-table><connection/>
<schema/>
<table/>
<size_limit_lines/>
<interval/>
<timeout_days/>
<field><id>ID_BATCH</id><enabled>Y</enabled><name>ID_BATCH</name></field><field><id>CHANNEL_ID</id><enabled>Y</enabled><name>CHANNEL_ID</name></field><field><id>TRANSNAME</id><enabled>Y</enabled><name>TRANSNAME</name></field><field><id>STATUS</id><enabled>Y</enabled><name>STATUS</name></field><field><id>LINES_READ</id><enabled>Y</enabled><name>LINES_READ</name><subject/></field><field><id>LINES_WRITTEN</id><enabled>Y</enabled><name>LINES_WRITTEN</name><subject/></field><field><id>LINES_UPDATED</id><enabled>Y</enabled><name>LINES_UPDATED</name><subject/></field><field><id>LINES_INPUT</id><enabled>Y</enabled><name>LINES_INPUT</name><subject/></field><field><id>LINES_OUTPUT</id><enabled>Y</enabled><name>LINES_OUTPUT</name><subject/></field><field><id>LINES_REJECTED</id><enabled>Y</enabled><name>LINES_REJECTED</name><subject/></field><field><id>ERRORS</id><enabled>Y</enabled><name>ERRORS</name></field><field><id>STARTDATE</id><enabled>Y</enabled><name>STARTDATE</name></field><field><id>ENDDATE</id><enabled>Y</enabled><name>ENDDATE</name></field><field><id>LOGDATE</id><enabled>Y</enabled><name>LOGDATE</name></field><field><id>DEPDATE</id><enabled>Y</enabled><name>DEPDATE</name></field><field><id>REPLAYDATE</id><enabled>Y</enabled><name>REPLAYDATE</name></field><field><id>LOG_FIELD</id><enabled>Y</enabled><name>LOG_FIELD</name></field></trans-log-table>
<perf-log-table><connection/>
<schema/>
<table/>
<interval/>
<timeout_days/>
<field><id>ID_BATCH</id><enabled>Y</enabled><name>ID_BATCH</name></field><field><id>SEQ_NR</id><enabled>Y</enabled><name>SEQ_NR</name></field><field><id>LOGDATE</id><enabled>Y</enabled><name>LOGDATE</name></field><field><id>TRANSNAME</id><enabled>Y</enabled><name>TRANSNAME</name></field><field><id>STEPNAME</id><enabled>Y</enabled><name>STEPNAME</name></field><field><id>STEP_COPY</id><enabled>Y</enabled><name>STEP_COPY</name></field><field><id>LINES_READ</id><enabled>Y</enabled><name>LINES_READ</name></field><field><id>LINES_WRITTEN</id><enabled>Y</enabled><name>LINES_WRITTEN</name></field><field><id>LINES_UPDATED</id><enabled>Y</enabled><name>LINES_UPDATED</name></field><field><id>LINES_INPUT</id><enabled>Y</enabled><name>LINES_INPUT</name></field><field><id>LINES_OUTPUT</id><enabled>Y</enabled><name>LINES_OUTPUT</name></field><field><id>LINES_REJECTED</id><enabled>Y</enabled><name>LINES_REJECTED</name></field><field><id>ERRORS</id><enabled>Y</enabled><name>ERRORS</name></field><field><id>INPUT_BUFFER_ROWS</id><enabled>Y</enabled><name>INPUT_BUFFER_ROWS</name></field><field><id>OUTPUT_BUFFER_ROWS</id><enabled>Y</enabled><name>OUTPUT_BUFFER_ROWS</name></field></perf-log-table>
<channel-log-table><connection/>
<schema/>
<table/>
<timeout_days/>
<field><id>ID_BATCH</id><enabled>Y</enabled><name>ID_BATCH</name></field><field><id>CHANNEL_ID</id><enabled>Y</enabled><name>CHANNEL_ID</name></field><field><id>LOG_DATE</id><enabled>Y</enabled><name>LOG_DATE</name></field><field><id>LOGGING_OBJECT_TYPE</id><enabled>Y</enabled><name>LOGGING_OBJECT_TYPE</name></field><field><id>OBJECT_NAME</id><enabled>Y</enabled><name>OBJECT_NAME</name></field><field><id>OBJECT_COPY</id><enabled>Y</enabled><name>OBJECT_COPY</name></field><field><id>REPOSITORY_DIRECTORY</id><enabled>Y</enabled><name>REPOSITORY_DIRECTORY</name></field><field><id>FILENAME</id><enabled>Y</enabled><name>FILENAME</name></field><field><id>OBJECT_ID</id><enabled>Y</enabled><name>OBJECT_ID</name></field><field><id>OBJECT_REVISION</id><enabled>Y</enabled><name>OBJECT_REVISION</name></field><field><id>PARENT_CHANNEL_ID</id><enabled>Y</enabled><name>PARENT_CHANNEL_ID</name></field><field><id>ROOT_CHANNEL_ID</id><enabled>Y</enabled><name>ROOT_CHANNEL_ID</name></field></channel-log-table>
<step-log-table><connection/>
<schema/>
<table/>
<timeout_days/>
<field><id>ID_BATCH</id><enabled>Y</enabled><name>ID_BATCH</name></field><field><id>CHANNEL_ID</id><enabled>Y</enabled><name>CHANNEL_ID</name></field><field><id>LOG_DATE</id><enabled>Y</enabled><name>LOG_DATE</name></field><field><id>TRANSNAME</id><enabled>Y</enabled><name>TRANSNAME</name></field><field><id>STEPNAME</id><enabled>Y</enabled><name>STEPNAME</name></field><field><id>STEP_COPY</id><enabled>Y</enabled><name>STEP_COPY</name></field><field><id>LINES_READ</id><enabled>Y</enabled><name>LINES_READ</name></field><field><id>LINES_WRITTEN</id><enabled>Y</enabled><name>LINES_WRITTEN</name></field><field><id>LINES_UPDATED</id><enabled>Y</enabled><name>LINES_UPDATED</name></field><field><id>LINES_INPUT</id><enabled>Y</enabled><name>LINES_INPUT</name></field><field><id>LINES_OUTPUT</id><enabled>Y</enabled><name>LINES_OUTPUT</name></field><field><id>LINES_REJECTED</id><enabled>Y</enabled><name>LINES_REJECTED</name></field><field><id>ERRORS</id><enabled>Y</enabled><name>ERRORS</name></field><field><id>LOG_FIELD</id><enabled>N</enabled><name>LOG_FIELD</name></field></step-log-table>
    </log>
    <maxdate>
      <connection/>
      <table/>
      <field/>
      <offset>0.0</offset>
      <maxdiff>0.0</maxdiff>
    </maxdate>
    <size_rowset>10000</size_rowset>
    <sleep_time_empty>50</sleep_time_empty>
    <sleep_time_full>50</sleep_time_full>
    <unique_connections>N</unique_connections>
    <feedback_shown>Y</feedback_shown>
    <feedback_size>50000</feedback_size>
    <using_thread_priorities>Y</using_thread_priorities>
    <shared_objects_file/>
    <capture_step_performance>N</capture_step_performance>
    <step_performance_capturing_delay>1000</step_performance_capturing_delay>
    <step_performance_capturing_size_limit>100</step_performance_capturing_size_limit>
    <dependencies>
    </dependencies>
    <partitionschemas>
    </partitionschemas>
    <slaveservers>
    </slaveservers>
    <clusterschemas>
    </clusterschemas>
  <created_user>-</created_user>
  <created_date>2012/11/16 13:59:51.117</created_date>
  <modified_user>-</modified_user>
  <modified_date>2014/05/12 15:22:12.008</modified_date>
  </info>
  <notepads>
  </notepads>
  <connection>
    <name>backupConn</name>
    <server>${CLIENT_DATABASE_IP}</server>
    <type>ORACLE</type>
    <access>Native</access>
    <database>${CLIENT_DATABASE_NAME}</database>
    <port>${CLIENT_DATABASE_PORT}</port>
    <username>${CLIENT_DATABASE_USERNAME}</username>
    <password>${CLIENT_DATABASE_PASSWORD}</password>
    <servername/>
    <data_tablespace/>
    <index_tablespace/>
    <attributes>
      <attribute><code>FORCE_IDENTIFIERS_TO_LOWERCASE</code><attribute>N</attribute></attribute>
      <attribute><code>FORCE_IDENTIFIERS_TO_UPPERCASE</code><attribute>N</attribute></attribute>
      <attribute><code>IS_CLUSTERED</code><attribute>N</attribute></attribute>
      <attribute><code>PORT_NUMBER</code><attribute>${CLIENT_DATABASE_PORT}</attribute></attribute>
      <attribute><code>QUOTE_ALL_FIELDS</code><attribute>N</attribute></attribute>
      <attribute><code>SUPPORTS_BOOLEAN_DATA_TYPE</code><attribute>N</attribute></attribute>
      <attribute><code>USE_POOLING</code><attribute>N</attribute></attribute>
    </attributes>
  </connection>
  <connection>
    <name>serverConn</name>
    <server>${SERVER_DATABASE_IP}</server>
    <type>ORACLE</type>
    <access>Native</access>
    <database>${SERVER_DATABASE_NAME}</database>
    <port>${SERVER_DATABASE_PORT}</port>
    <username>${SERVER_DATABASE_USERNAME}</username>
    <password>${SERVER_DATABASE_PASSWORD}</password>
    <servername/>
    <data_tablespace/>
    <index_tablespace/>
    <attributes>
      <attribute><code>FORCE_IDENTIFIERS_TO_LOWERCASE</code><attribute>N</attribute></attribute>
      <attribute><code>FORCE_IDENTIFIERS_TO_UPPERCASE</code><attribute>N</attribute></attribute>
      <attribute><code>IS_CLUSTERED</code><attribute>N</attribute></attribute>
      <attribute><code>PORT_NUMBER</code><attribute>${SERVER_DATABASE_PORT}</attribute></attribute>
      <attribute><code>QUOTE_ALL_FIELDS</code><attribute>N</attribute></attribute>
      <attribute><code>SUPPORTS_BOOLEAN_DATA_TYPE</code><attribute>N</attribute></attribute>
      <attribute><code>USE_POOLING</code><attribute>N</attribute></attribute>
    </attributes>
  </connection>
  <connection>
    <name>wardConn</name>
    <server>${CLIENT_DATABASE_IP}</server>
    <type>ORACLE</type>
    <access>Native</access>
    <database>${CLIENT_DATABASE_NAME}</database>
    <port>${CLIENT_DATABASE_PORT}</port>
    <username>${CLIENT_DATABASE_USERNAME}</username>
    <password>${CLIENT_DATABASE_PASSWORD}</password>
    <servername/>
    <data_tablespace/>
    <index_tablespace/>
    <attributes>
      <attribute><code>FORCE_IDENTIFIERS_TO_LOWERCASE</code><attribute>N</attribute></attribute>
      <attribute><code>FORCE_IDENTIFIERS_TO_UPPERCASE</code><attribute>N</attribute></attribute>
      <attribute><code>INITIAL_POOL_SIZE</code><attribute>50</attribute></attribute>
      <attribute><code>IS_CLUSTERED</code><attribute>N</attribute></attribute>
      <attribute><code>MAXIMUM_POOL_SIZE</code><attribute>500</attribute></attribute>
      <attribute><code>PORT_NUMBER</code><attribute>${CLIENT_DATABASE_PORT}</attribute></attribute>
      <attribute><code>QUOTE_ALL_FIELDS</code><attribute>N</attribute></attribute>
      <attribute><code>SUPPORTS_BOOLEAN_DATA_TYPE</code><attribute>N</attribute></attribute>
      <attribute><code>USE_POOLING</code><attribute>Y</attribute></attribute>
    </attributes>
  </connection>
  <order>
  <hop> <from>增加常量</from><to>增加常量 2</to><enabled>Y</enabled> </hop>  <hop> <from>计算器</from><to>Set Variables 2</to><enabled>Y</enabled> </hop>  <hop> <from>调用DB存储过程</from><to>调用DB存储过程 2</to><enabled>Y</enabled> </hop>  <hop> <from>调用DB存储过程 2</from><to>增加常量</to><enabled>Y</enabled> </hop>  <hop> <from>增加常量 2</from><to>计算器</to><enabled>Y</enabled> </hop>  </order>
  <step>
    <name>Set Variables 2</name>
    <type>SetVariable</type>
    <description/>
    <distribute>Y</distribute>
    <copies>1</copies>
         <partitioning>
           <method>none</method>
           <schema_name/>
           </partitioning>
    <fields>
      <field>
        <field_name>TIMEDIFF</field_name>
        <variable_name>DOWNLOAD_SERVER_COMMON_DATA_LEAD</variable_name>
        <variable_type>JVM</variable_type>
        <default_value/>
        </field>
      </fields>
    <use_formatting>Y</use_formatting>
     <cluster_schema/>
 <remotesteps>   <input>   </input>   <output>   </output> </remotesteps>    <GUI>
      <xloc>118</xloc>
      <yloc>273</yloc>
      <draw>Y</draw>
      </GUI>
    </step>

  <step>
    <name>增加常量</name>
    <type>Constant</type>
    <description/>
    <distribute>Y</distribute>
    <copies>1</copies>
         <partitioning>
           <method>none</method>
           <schema_name/>
           </partitioning>
    <fields>
      <field>
        <name>DIVIDEND</name>
        <type>Integer</type>
        <format/>
        <currency/>
        <decimal/>
        <group/>
        <nullif>86400000</nullif>
        <length>-1</length>
        <precision>-1</precision>
      </field>
    </fields>
     <cluster_schema/>
 <remotesteps>   <input>   </input>   <output>   </output> </remotesteps>    <GUI>
      <xloc>388</xloc>
      <yloc>241</yloc>
      <draw>Y</draw>
      </GUI>
    </step>

  <step>
    <name>增加常量 2</name>
    <type>Constant</type>
    <description/>
    <distribute>Y</distribute>
    <copies>1</copies>
         <partitioning>
           <method>none</method>
           <schema_name/>
           </partitioning>
    <fields>
      <field>
        <name>DEVIATION</name>
        <type>Integer</type>
        <format/>
        <currency/>
        <decimal/>
        <group/>
        <nullif>10000</nullif>
        <length>-1</length>
        <precision>-1</precision>
      </field>
    </fields>
     <cluster_schema/>
 <remotesteps>   <input>   </input>   <output>   </output> </remotesteps>    <GUI>
      <xloc>364</xloc>
      <yloc>362</yloc>
      <draw>Y</draw>
      </GUI>
    </step>

  <step>
    <name>计算器</name>
    <type>Calculator</type>
    <description/>
    <distribute>Y</distribute>
    <copies>1</copies>
         <partitioning>
           <method>none</method>
           <schema_name/>
           </partitioning>
       <calculation><field_name>TEMP</field_name>
<calc_type>SUBTRACT</calc_type>
<field_a>LOCAL_DATE</field_a>
<field_b>SERVER_DATE</field_b>
<field_c/>
<value_type>BigNumber</value_type>
<value_length>-1</value_length>
<value_precision>-1</value_precision>
<remove>N</remove>
<conversion_mask/>
<decimal_symbol/>
<grouping_symbol/>
<currency_symbol/>
</calculation>
       <calculation><field_name>TEMP2</field_name>
<calc_type>ADD</calc_type>
<field_a>TEMP</field_a>
<field_b>DEVIATION</field_b>
<field_c/>
<value_type>BigNumber</value_type>
<value_length>-1</value_length>
<value_precision>-1</value_precision>
<remove>N</remove>
<conversion_mask/>
<decimal_symbol/>
<grouping_symbol/>
<currency_symbol/>
</calculation>
       <calculation><field_name>TIMEDIFF</field_name>
<calc_type>DIVIDE</calc_type>
<field_a>TEMP2</field_a>
<field_b>DIVIDEND</field_b>
<field_c/>
<value_type>BigNumber</value_type>
<value_length>-1</value_length>
<value_precision>-1</value_precision>
<remove>N</remove>
<conversion_mask/>
<decimal_symbol/>
<grouping_symbol/>
<currency_symbol/>
</calculation>
     <cluster_schema/>
 <remotesteps>   <input>   </input>   <output>   </output> </remotesteps>    <GUI>
      <xloc>251</xloc>
      <yloc>272</yloc>
      <draw>Y</draw>
      </GUI>
    </step>

  <step>
    <name>调用DB存储过程</name>
    <type>DBProc</type>
    <description/>
    <distribute>Y</distribute>
    <copies>1</copies>
         <partitioning>
           <method>none</method>
           <schema_name/>
           </partitioning>
    <connection>serverConn</connection>
    <procedure>F_GET_SYSDATE2INT</procedure>
    <lookup>
    </lookup>
    <result>
      <name>SERVER_DATE</name>
      <type>BigNumber</type>
    </result>
    <auto_commit>Y</auto_commit>
     <cluster_schema/>
 <remotesteps>   <input>   </input>   <output>   </output> </remotesteps>    <GUI>
      <xloc>140</xloc>
      <yloc>134</yloc>
      <draw>Y</draw>
      </GUI>
    </step>

  <step>
    <name>调用DB存储过程 2</name>
    <type>DBProc</type>
    <description/>
    <distribute>Y</distribute>
    <copies>1</copies>
         <partitioning>
           <method>none</method>
           <schema_name/>
           </partitioning>
    <connection>wardConn</connection>
    <procedure>F_GET_SYSDATE2INT</procedure>
    <lookup>
    </lookup>
    <result>
      <name>LOCAL_DATE</name>
      <type>BigNumber</type>
    </result>
    <auto_commit>Y</auto_commit>
     <cluster_schema/>
 <remotesteps>   <input>   </input>   <output>   </output> </remotesteps>    <GUI>
      <xloc>353</xloc>
      <yloc>132</yloc>
      <draw>Y</draw>
      </GUI>
    </step>

  <step_error_handling>
  </step_error_handling>
   <slave-step-copy-partition-distribution>
</slave-step-copy-partition-distribution>
   <slave_transformation>N</slave_transformation>
</transformation>

2、通过NTP方式校对系统时间

1）如果可以连接到Internet网中，则可以采用国际上的校时服务器去校对时间

2）如果在内网中且不能上网，则可以自己定义一台都可访问到的台式机作为NTP服务器，其他机子与该计算机教士即可，下面是window NTP校时服务器的设置。

可以参考文章：NTP服务同步时间方案