1.3 Trans Processing

• EDI:

  • Standard for Electronic data interchange
• EDI:

  • Must be in standard format so that the computer can understand it
  • Replaces post,email therefore saving paper and helping the environment
• Transaction processing:

  • Any information processing that is divided into individual indivisible operations called transactions
• The base functionality of all relational database is referred to the acronym CRUD:

  • Create
  • Read(Select)
  • Update
  • Delete
• ACID:

  • Set of properties to ensure data integrity is maintained
  • As well as transactions are processed reliably
• ACID:

  • Atomicity
  • Consistency
  • Isolation
  • Durability
• Atomicity:

  • Ensures that a transaction is either completely performed or not at all
  • A half-completed change must not be saved back to the database
• Consistency:

  • Ensures that no transactions can violate any of the defined validation rules.
  • Referential integrity, will always be upheld, meaning any change in the database must remain the overall state of the database
  • For example, money debited from one account must be balanced by the money being credited in another
• Isolation:

  • A transaction must not be interrupted by another transaction
  • The transaction must occur in isolation so other users or process cannot access the data concerned
• Durability:

  • Once a transaction is committed, it will remain even in the event of a system failure.
  • DBMS writes the effects of the transactions into secondary storage rather than RAM.
  • Held in buffer until all elements are completed, only the changes will to the database tables be made
• Record locking:

  • Prevents simultaneous access to objects in a database to prevent updates or being lost or inconsistencies in the data arising
• Disadvantages of record locking:

  • A deadlock can arise where if 2 users are attempting to update 2 records at the same time
  • Can cause delays (as users wait for access)
• Serialisation:

  • Ensures that transactions do not overlap in time and therefore cannot interfere with each other or lead to updates being lost
• Timestamp ordering:

  • If 2 transactions affect the same object, the transaction with the earlier timestamp should be applied first
• Timestamp ordering works by:

  • Every object in the database has a read and write timestamp
  • Which is updated whenever an object is read or written
  • When the user tries to save an update, if the read timestamp is not the same as it was when they started the transaction, the DMBS knows another user has accessed the same object.
• Commitment ordering:

  • Another serialisation technique to ensure that no transactions are lost if 2 clients are simultaneously trying to update record
• Commitment ordering works by:

  • Transactions are ordered in terms of their dependencies on one another as well as the time they were initiated.
  • Can be used to prevent deadlock by blocking one request until another is completed
• In built redundancy protects transaction data from system failure:

  • Duplicate hardware is located in different geographical areas
  • Mirrors every transaction that takes place
  • If this fails, the backup system takes over
• Ways to validate data:

  • Range check…
  • Length check…
  • Type check..
  • Format check ..