
7-4 Test: Ensuring Stability and Performance
Developing Fallback Procedures Adaptive Server Enterprise 11.5
- For an around-the-clock operation, you may decide that an
initial goal of breaking even is reasonable.
• For cutover without replication, gear the migration for
equivalent performance between the old and new systems. A
goal of breaking even the first week is reasonable.
•Aphased cutover is subject to the highest performance
expectations. Some performance tuning of the production
workload after cutover of the production server may be best. You
can time the production server cutover to occur as soon as
performance gains are acceptable and testing is successful.
See Chapter 4, “Prepare: Writing a Plan and Getting Ready to
Migrate” for information on migration methods.
Developing Fallback Procedures
Before you begin testing, be sure that you know how you will return
the test system to a known state when you have a problem. You will
use the same fallback plan when you migrate the production system.
During simple testing cycles, it’s sometimes faster to back out
unwanted changes. However, this is not recommended for timed
runs during benchmarking. You have to restore from backup after
each timed run to return the system to a known state.
Back up all databases before and after the test system upgrade, as
you would for a “real” upgrade. The backups preserve the layout of
data on disk and help you avoid confusion due to fragmentation and
page splits.
To ensure source code control, use scripts for all changes to objects.
This makes it much easier to recreate your environment if necessary.
Summary of Testing Techniques
You can use a variety of testing techniques in your test plan. The
following table describes the advantages and disadvantages of
various testing techniques and tools, including:
• Ad-hoc testing
• Manual test scripts and cases
• Keystroke capture tools
• Transaction generators
Commenti su questo manuale