【精品】建设工程竣工验收消防备案表 title60_第1页
【精品】建设工程竣工验收消防备案表 title60_第2页
【精品】建设工程竣工验收消防备案表 title60_第3页
【精品】建设工程竣工验收消防备案表 title60_第4页
【精品】建设工程竣工验收消防备案表 title60_第5页
已阅读5页,还剩15页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

RACInternals,Julian DykeIndependent Consultant,Web Version,, 2007 Julian Dyke,2,Agenda,Transactions in RACCross Instance Consistent Reads,3,Introduction,4,System Change Number,In RAC clusters SCN must be maintained across all nodes in clusterSCN propagation scheme differs according to versionIn Oracle 9.2 and below defaults to Lamport algorithmLamport or SCN Scheme 2 in alert.logSCN piggy-backed on GCS/GES messagesRecorded in redo logDefault delay of 7 secondsIn Oracle 10.1 and above uses a new algorithmSCN Scheme 3 in alert.log Broadcast on commitApparently no delay,5,MAX_COMMIT_PROPAGATION_DELAY,Prior to Oracle 10.2 Default value is 700 centiseconds (7 seconds)Specifies maximum time taken for a COMMIT on one node to be reflected on other nodes in the clusterFor some applications, value must be set to 0 (Broadcast on commit) including:E-Business suiteSAPIn Oracle 10.2 and above default value is 0,6,LMS Background Processes,LMS background processes:Implement cache fusionServe both consistent and current versions of blocks in cache of local instance to other instancesMaintain local part of Global Resource DirectoryMinimum of 1 LMS process per instanceMaximum is version dependentOracle 9.210Oracle 10.120Oracle 10.236Prior to Oracle 10.1, could be configured using _lm_lms parameterIn Oracle 10.1 and above, initial number of LMS processes specified by gcs_server_processes parameter,7,LMS Background Processes,Each LMS background process manages a set of blocksDetermined by hash function based on number of LMS background processesConsequently a block will always be handled by the same LMS processNumber of blocks served recorded in Session / System statisticsV$CR_BLOCK_SERVER,8,Cross Instance Consistent Read,uba 800777.530.12,uba: -,12,13,14,Undo Header,Data Block 42,Undo Block 800777,Session 27,LMS0,Instance 2,Instance 1,Session 15,Data Block 42 (copy),9,V$CR_BLOCK_SERVER,10,Light Works Rule,In theory, once a block has been written to disk, the LMS process will not attempt to read it again when responding to a consistent read requestLight Works RulePrevents LMS processes from going to disk when responding to CR requests for data, undo or undo segment blocksCan prevent LMS process from completing its response to a CR request,11,GC Read Committed Block,Instance 1,Instance 2,Block 42,UndoBlock,Committed Block - Data Block on disk,12,GC Read Committed Block,Instance 1,Instance 2,Block 42,UndoBlock,Committed Block - Data Block in buffer cache,13,GC Read Uncommitted Block,Uncommitted changes MUST be flushed to the redo log before the LMS process can ship a consistent block to another instanceReading process must wait until redo log changes have been written to redo log by LMS processBad for standard RAC databasesReads must wait for redo log writesWorse for extended / stretch RAC clustersIncreased latency of cross site disk communications,14,GC Read Uncommitted Block,For each block on which a consistent read is performed, a redo log flush must first be performedNumber of redo log flushes is recorded in the FLUSHES column of V$CR_BLOCK_SERVERRedo log flush time is recorded in the gc cr block flush time statistic for the LMS processwill increase time taken to serve consistent blockwill increase time taken to perform consistent readIf LMS processes become very busy, consistent reads will experience high wait times e.g. for a full table scan gc cr multi block request,15,GC Read Uncommitted Block,Instance 1,Instance 2,Block 42,UndoBlock,Block 42Copy,Uncommitted Block - Data Block in buffer cache,16,GC Read Uncommitted Block,Instance 1,Instance 2,Block 42,UndoBlock,Uncommitted Block - Data Block on disk,17,Consistent Reads in RAC,If possible, blocks will always be read from the cache of another instanceUndo blocks will be flushed to disk more frequently when:All columns are updatedIndexed columns are updatedSingle rows inserted as opposed to using array insertsTransactions are regularly rolled back Rows locked using SELECT FOR UPDATEData blocks will be flushed to disk more frequently when:Most transactions are read-only,18,Consistent Reads in RAC,Consistent read response times in RAC can be reduced by:Avoid reading uncommitted blocks on remote nodesPartitioningLimiting number of rows per block Specifying SCNMinimizing size of transactions on remote nodesMust retain ACID propertiesMay be possible to use application logic to synchronize writes and readsIncreasing number of LMS processes on remote nodeShould be added dynamically by kernelAlso by obvious hardware changes

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

最新文档

评论

0/150

提交评论