Standards of Exchange Administration.


116 views
Uploaded on:
Description
Standards of Exchange Administration. Plot. Exchange ideas and conventions Execution effect of simultaneousness control Execution tuning. Application Developer (e.g., business investigator, Information planner). Application. Modern Application Software engineer (e.g., SAP administrator).
Transcripts
Slide 1

Standards of Transaction Management

Slide 2

Outline Transaction ideas & conventions Performance effect of simultaneousness control Performance tuning

Slide 3

Application Programmer (e.g., business examiner, Data modeler) Application Sophisticated Application Programmer (e.g., SAP administrator) Query Processor Indexes Storage Subsystem Concurrency Control Recovery DBA, Tuner Operating System Hardware [Processor(s), Disk(s), Memory]

Slide 4

Transaction Concepts & Protocols Transaction An intelligent unit of database handling A succession of start, peruses/composes, end Unit of recuperation, consistency, simultaneousness Transaction Processing Systems Large databases with various clients executing database exchanges Examples Banking frameworks, carrier reservations, general store checkouts, ...

Slide 5

Transition STATE Transaction States read-thing, compose thing start exchange end-exchange confer Active Partially Committed prematurely end prematurely end Failed Terminated

Slide 6

Interleaved Transactions An and B are simultaneous exchanges An A B Time t1 t2 t3 t4 t5

Slide 7

Transaction “ Correctness ” ACID properties Atomicity Consistency Isolation Durability Enforced by simultaneousness control and recuperation routines for the DBMS

Slide 8

Serial Schedule An arrangement of read & compose operations from different exchanges R1[X] W3[Y] R2[X] W2[Y] W1[X] W2[X] Serial calendar No interleaved operations from the partaking exchanges W3[Z] R3[Y] R1[X] W1[Y] R2[Y] W2[Z] W2[X] Always adjust, yet … so moderate! A timetable that is equal to some serial calendar is right as well

Slide 9

Serializable Schedule T1 T2 R( A ) W( A ) R( A ) W( A ) R( B ) W( B ) R( B ) W( B ) Commit

Slide 10

Equivalent Schedules 2 calendars are comparable if the exchanges Read the same qualities Produce the same yield Have the same impact on the database Examples R1[X] W2[X] R3[Y] W1[Y] R2[Y] W3[Z] W2[Z] W3[Z] R3[Y] R1[X] W1[Y] R2[Y] W2[Z] W2[X] R1[X] W1[Y] R2[Y] W3[Z] W2[Z] R3[Y] 1 and 2 are identical; not 3

Slide 11

Serializable Schedule Theorem A calendar is serializable if there is a serial calendar such that for each clashing pair of operations, the two operations show up in the same request in both timetables. 2 operations strife in the event that they are on the same item and one is a compose Example 1 is serializable

Slide 12

WR Conflicts T1 T2 T1 exchange $100 from A to B, and T2 increases both and B by 6% (An and B have $200 at first) R( A ) ( $200 ) W( A ) ( $100 ) R( A ) ( 100 ) W( A ) ( 106 ) Dirty read R( B ) ( 200 ) W( B ) ( 212 ) Commit R(A) R( B ) ( 212 ) W( B ) ( 312 ) Unrepeatable Read (UR) Commit

Slide 13

WW Conflicts T1 T2 T1 to set both An and B to $1000, T2 to set both An and B to $2000 R( A ) W( A ) ( $1000 ) R( B ) W( B ) ( $2000) R( B ) W( B ) ( $1000 ) R( A ) W( A ) ( $2000 ) Commit Lost Update! Confer

Slide 14

Concurrency Control Enforces Serializability Most business DBMS use conventions (an arrangement of tenets) which when implemented by DBMS guarantee the serializability of all calendars in which exchanges take an interest. Serializability testing after execution is useless; how to redress? This done by Concurrency Control

Slide 15

Concurrency Control Protocols Commercially acknowledged components Locking Timestamps Others instruments Multi-adaptation and idealistic conventions Granularity issues

Slide 16

Locking is utilized to synchronize gets to by simultaneous exchanges on information things An idea additionally found in working frameworks and simultaneous programming A lock is a variable for an information thing, that depicts the thing\'s status concerning admissible operations

Slide 17

Types of Locks Binary locks Locked, or Unlocked Check before enter; hold up when bolted; lock after enter; open after utilization (and wakeup one holding up exchange). Straightforward however excessively prohibitive Read/Write secures business DBMS read-bolted compose bolted Unlocked R-lock W-lock R-lock Y N W-lock N

Slide 18

Read/Write Locking Scheme An exchange T must issue read-lock (X) or compose lock before any read-thing (X) T must issue compose lock (X) before any compose thing (X) T must issue open thing (X) in the wake of finishing all read-thing (X) and compose thing (X) T won\'t issue a read-lock (X) if T as of now holds a read/compose lock on X T won\'t issue compose lock (X) if T as of now holds a compose lock on X

Slide 19

Does Locking Ensure Serializability? T1 read-lock (Y); read-thing (Y); open (Y); compose lock (X); read-thing (X); X:=X+Y; compose thing (X); open (X); T2 read-lock (X); read-thing (X); open (X); compose lock (Y); read-thing (Y); Y:=X+Y; compose thing (Y); open (Y); X opened too soon Y opened too soon X == Y (orignal X + originalY) For serializable T1T2, X == X + Y == 2Y + originalX? Can\'t serialize T1 and T2

Slide 20

Need for Locking Protocol Locking alone does not guarantee serializability! We require a locking convention An arrangement of guidelines that direct the situating of locking and opening operations, along these lines ensuring serializability

Slide 21

Two-Phase Locking (2PL) An exchange takes after the two-stage convention if every single locking operation go before the first opening operation read-lock (X) compose lock (X) compose lock (Y) read-lock (Y) open (X) open (Y) Phase 1: Growing Phase 2: Shrinking

Slide 22

2PL Variants Basic 2PL Conservative 2PL Locking operations go before exchange execution Make beyond any doubt can procure fundamental locks Strict 2PL Unlocking of compose locks after submit (or prematurely end) Avoid falling prematurely end Rigorous 2PL Unlocking of all locks after confer (or prematurely end)

Slide 23

Limitations of 2PL Some serializable calendars may not be allowed Performance not ideal 2PL (and securing general) may bring about gridlocks and starvation Deadlock: no exchanges can continue Starvation: some exchange hold up perpetually

Slide 24

Lock Granularity Larger size - lower simultaneousness Smaller size - higher overhead What is the best thing size? Handling a blend of exchanges? Relies on upon the kind of exchanges Multiple granularity locking plan, changing the information\'s span thing progressively

Slide 25

Performance of Locking Throughput Thrashing # of Active Transactions Overhead: blocking Increasing the throughput: Locking littler size items Reducing locking time Reducing problem areas

Slide 26

Other CC Protocols Timestamp based Multi-form based Optimistic simultaneousness control No checking is done before or amid exchange execution The exchange is accepted toward the end of execution, by checking if serializability has been damaged

Slide 27

Summary of Transaction Concepts Baseline: Serial Schedule Transaction Correctness  Other CC Protocols Timestamp Multi-rendition Optimistic Strict 2PL A C I D   2PL  Ideal: Serializable Schedule

Slide 28

Summary To enhance execution Interleave exchanges Correctness: ACID Serial timetable is right Serializable calendar is proportionate to some serial timetable Concurrency control authorizes serializability 2PL Deadlock Starvation Granularity Optimistic Timestamping Multi-adaptation

Slide 29

Performance Impact of Concurrency Control Lock dispute Deadlock

Slide 30

Performance Impact of Concurrency Control LONG exchanges are punished .:tsli

Recommended
View more...