Databases Reference
In-Depth Information
Well-defined performance requirements
OLTP systems are central to core business operations, so users must be able to
depend on a consistent response time. OLTP systems often involve service level
agreements that state the expected response times.
High availability
These systems are typically deemed mission-critical with significant costs resulting
from downtime.
Scalability
The ability to increase transaction volumes without significant degradation in per‐
formance allows OLTP systems to handle fluctuations in business activity.
In short, OLTP systems must be able to deliver consistent performance at any time,
regardless of system load. Anything that affects these core systems can produce a ripple
effect throughout your entire organization, affecting both revenue and profitability.
Online and batch transaction processing
Online transaction processing implies direct and conversational interaction between
the transaction processing system and its users. Users enter and query data using forms
that interact with the backend database. Editing and validation of data occur at the time
the transactions are submitted by users.
Batch processing occurs without user interaction. Batches of transactions are fed from
source files to the operational system. Errors are typically reported in exception files or
logs and are reviewed by users or operators later on. Virtually all OLTP systems have a
batch component: jobs that can execute in off-peak hours for reporting, payroll runs,
posting of accounting entries, and so on.
Many large companies have batch-oriented mainframe systems that are so thoroughly
embedded in the corporate infrastructure that they cannot be replaced or removed. A
common practice is to “frontend” these legacy systems with OLTP systems that provide
more modern interfaces. Users interact with the OLTP system to enter transactions.
Batch files are extracted from the OLTP system and fed into the downstream legacy
applications.
Once the batch processing is done, extracts are produced from the batch systems and
are used to refresh the OLTP systems. This extraction process provides the users with
a more sophisticated interface with online validation and editing, but it preserves the
flow of data through the entrenched batch systems. While this process seems costly, it's
typically more attractive than the major surgery that would replace older systems. To
compound the difficulty, in some cases the documentation of these older systems is
incomplete and the employees who understand the inner workings have retired or
moved on.
