Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 1Crash RecoveryDatabase Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 2Review: The ACID propertiesAA tomicity: All actions in a Xact happen, or none happenCC onsistency: Each Xact transforms the database from one consistent state to anotherII solation: Execution of concurrent transactions is as though they are evaluated in some serial orderD D urability: If a Xact commits, its effects persist The Recovery Manager guarantees Atomicity & Durability.Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 3Motivation Atomicity: – Transactions may abort (“Rollback”). Durability:– What if DBMS stops running? (Causes?)crash! Desired Behavior after system restarts:– T1, T2 & T3 should be durable.– T4 & T5 should be aborted (effects not seen).T1T2T3T4T5Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 4Assumptions Concurrency control is in effect – Strict 2PL, in particular. Updates are happening “in place”– i.e. data is overwritten on (deleted from) the disk. A simple scheme to guarantee Atomicity & Durability?Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 5Handling the Buffer Pool Force every write to disk?– Poor response time.– But provides durability. Steal buffer-pool frames from uncommited Xacts?– If not, poor throughput.– If so, how can we ensure atomicity?ForceNo ForceNo Steal StealTrivial (?)DesiredDatabase Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 6More on Steal and Force STEAL (why enforcing Atomicity is hard)– To steal frame F: Current page in F (say P) is written to disk; some Xact holds lock on P. What if the Xact with the lock on P aborts? Must remember the old value of P at steal time (to support UNDOing the write to page P). NO FORCE (why enforcing Durability is hard)– What if system crashes before a modified page is written to disk?– Write as little as possible, in a convenient place, at commit time,to support REDOing modifications.Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 7Basic Idea: Logging Record REDO and UNDO information, for every update, in a log.– Sequential writes to log (put it on a separate disk).– Minimal info (diff) written to log, so multiple updates fit in a single log page. Log: An ordered list of REDO/UNDO actions– Log record contains: <XID, pageID, offset, length, old data, new data> – and additional control info (which we’ll see soon).Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 8Write-Ahead Logging (WAL) The Write-Ahead Logging Protocol: Must force the log record for an update before the corresponding data page gets to disk. Must write all log records for a Xact before commit. #1 guarantees Atomicity. #2 guarantees Durability. Exactly how is logging (and recovery!) done?– We’ll study the ARIES algorithms.Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 9WAL & the Log Each log record has a unique Log Sequence Number (LSN).– LSNs always increasing. Each data page contains a pageLSN.– The LSN of the most recent log record for an update to that page. System keeps track of flushedLSN.– The max LSN flushed so far. WAL: Before a page is written,– pageLSN ≤ flushedLSNLSNsDBpageLSNsRAMflushedLSNpageLSNLog recordsflushed to disk“Log tail”in RAMDatabase Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 10Log RecordsPossible log record types: Update Commit Abort End (signifies end of commit or abort) Compensation Log Records (CLRs)– for UNDO actionsprevLSNXIDtypelengthpageIDoffsetbefore-imageafter-imageLogRecord fields:updaterecordsonlyDatabase Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 11Other Log-Related State Transaction Table:– One entry per active Xact.– Contains XID, status (running/commited/aborted), and lastLSN. Dirty Page Table:– One entry per dirty page in buffer pool.– Contains recLSN -- the LSN of the log record which first caused the page to be dirty.Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 12Normal Execution of an Xact Series of reads & writes, followed by commit or abort.– We will assume that write is atomic on disk. In practice, additional details to deal with non-atomic writes. Strict 2PL.  STEAL, NO-FORCE buffer management, with Write-Ahead Logging.Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 13Checkpointing Periodically, the DBMS creates a checkpoint, in order to minimize the time taken to recover in the event of a system crash. Write to log:– begin_checkpoint record: Indicates when chkpt began.– end_checkpoint record: Contains current Xact table and dirty page table. This is a `fuzzy checkpoint’: Other Xacts continue to run; so these tables accurate only as of the time of the begin_checkpoint record. No attempt to force dirty pages to disk; effectiveness of checkpoint limited by oldest unwritten change to a dirty page. (So it’s a good idea to periodically flush dirty pages to disk!)– Store LSN of chkpt record in a safe place (master record).Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 14The Big Picture: What’s Stored WhereDBData pageseachwith apageLSNXact TablelastLSNstatusDirty Page TablerecLSNflushedLSNRAMprevLSNXIDtypelengthpageIDoffsetbefore-imageafter-imageLogRecordsLOGmaster recordDatabase Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 15Transaction Commit Write commit record to log. All log records up to Xact’s lastLSN are flushed.– Guarantees that flushedLSN ≥ lastLSN.– Note that log flushes are sequential, synchronous writes to disk.– Many log records per log page. Commit() returns. Write end record to log.Database Management Systems, 2ndEdition. R. Ramakrishnan and J. Gehrke 16Simple Transaction Abort For now, consider an explicit abort of a Xact.– No crash involved. We want to “play back” the log in reverse order, UNDOing updates.– Get lastLSN of Xact from Xact table.– Can follow chain of log records backward via the prevLSN field.– Before starting UNDO, write an Abort log record. For recovering from crash during UNDO!Database Management Systems, 2ndEdition. R. Ramakrishnan and J.