• Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint
Share this Page URL
Help

Chapter 17. Transaction Processing > Implementing Transaction Processing System...

Implementing Transaction Processing Systems

Of course, there is more to implementing a business application than ensuring ACID semantics, but a careful analysis of the transactional aspects provides a good basis for system design. In addition, taking this point of view early in the design is much easier than trying to retrofit an existing system to behave better.

Files

It is possible to build a complete transaction processing system on top of an ordinary filesystem. Indeed, in the early days of online transaction processing, systems were built from a TP monitor together with a filesystem. Computer filesystems perform a set of relatively simple operations: create, delete, open, close, read, write, and seek. Filesystems generally provide no assistance in structuring the contents of files and no assistance in managing transactions, except that some operating systems provide basic facilities to lock and unlock files, protecting them from concurrent access. (Locks are a way of providing isolation semantics at the risk of introducing performance problems and deadlocks.[4]) Nevertheless, files are valuable for several purposes.

[4] A deadlock occurs when two or more processes are waiting for resources and neither of them can make any progress. For example, suppose two transactions are in progress, and each needs resources A and B. If transaction 1 locks resource A and then attempts to lock resource B, while transaction 2 locks resource B and then attempts to lock resource A, neither transaction can proceed.


PREVIEW

                                                                          

Not a subscriber?

Start A Free Trial


  
  • Creative Edge
  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint