Download View File

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Open Database Connectivity wikipedia , lookup

Oracle Database wikipedia , lookup

IMDb wikipedia , lookup

Global serializability wikipedia , lookup

Ingres (database) wikipedia , lookup

Functional Database Model wikipedia , lookup

Relational model wikipedia , lookup

Database wikipedia , lookup

Commitment ordering wikipedia , lookup

Database model wikipedia , lookup

Microsoft Jet Database Engine wikipedia , lookup

Extensible Storage Engine wikipedia , lookup

Clusterpoint wikipedia , lookup

ContactPoint wikipedia , lookup

Serializability wikipedia , lookup

Concurrency control wikipedia , lookup

Transcript
Data and Database
Administration
Chapter 12 (Contd.)

Recovery and restart procedures
Transaction Integrity




Many DB transactions result in changes to one or
more DB records.If an error occurs during a
transaction processing, then some form of DB
recovery will be required.
To understand DB recovery, we must understand the
concept of transaction integrity.
A business transaction represents a business activity
which has a sequence of steps.
When processing transactions, DBMS must ensure
the four basic properties called ACID:

Atomic

meaning that the transaction can not be subdivided, it may be
processed in entirety or not at all.
Transaction Integrity (Contd.)

Consistent


Isolated


meaning that DB constraints that must be true before the
transactions must also be true after the transaction.
meaning changes to the DB are not revealed to the users until
the transaction is committed.
Durable

meaning that changes are permanent, once a transaction is
completed, no subsequent failures can revert back the effect
of transaction.
Recovery and Restart Procedures

Type of recovery procedure depends on




The nature of failure
The sophistication of DBMS recovery facilities
Operational policies and procedures
Techniques frequently used are:




Switch
Restore/run
Backward recovery
Forward recovery
Recovery and Restart Procedures: Switch




To switch to an existing copy of the database, the
database must be mirrored, i.e. two copies of the
database must be kept and updated simultaneously.
On failure, processing is switched to the duplicate
copy of the database.
This technique allows fastest recovery.
Level-1 RAID systems implement mirroring.


Instead of storing the DB on a large drive, it is it is
distributed across several smaller, less expensive
disks and mirrored to a second set of such disks.
On disk failure, system switches to the mirrored disks.
Recovery and Restart Procedures: Restore/run

This technique involves reprocessing the day’s transactions
(upto point of failure) against the backup copy of the
database/portion of DB.




Advantage


The database is shut down
then most recent copy of DB to be recovered is mounted,
and all transactions that have occurred since that copy are rerun.
simplicity (no need to create DB change journal, no restart
procedures)
Disadvantages




reprocessing overhead (depending on frequency of making backup
copies, several hours may be required)
Processing of new transactions will have to be deferred until recovery
is completed.
The sequence in which transactions are rerun, may lead to quite
different results.
Hence, restore/rerun is used only as a last resort in DB processing.
Backward Recovery (rollback)



Through this technique, the DBMS back outs or
undoes the unwanted changes to the database.
Before-images of the records are applied to the
database. Thus, the DB returns to its earlier
state.
Backward recovery is used to reverse changes
due to aborted or abnormally terminated
transactions.
Forward Recovery (rollforward)



A technique that starts with an earlier copy of the
database.
After-images (results of good transactions) are
applied to the database, and DB is quickly moved
forward to a later state.
It is much faster & accurate as compared to
restore/rerun because:


It does not involve the time consuming reprocessing of
transactions
Only the most recent after images need to applied. A DBrecord may have a series of after-images but the most
recent, good after-image is used in rollforward.
Types of DB failure

Four of the most common types of
failures/problems are:




Aborted transactions
Incorrect (but valid) data
System failure
Database loss or destruction