11g Database Upgrade Options
The next
several Blog entries will be centered on research that I conducted on the
possible upgrade options when moving the system to new hardware on both the
production and backup servers.
Today we briefly look a some of the options
available and I ranked them from 1-5 in several areas.
The simplest upgrade to perform is by installing 11g software
on the same server as the 10g instance resides and upgrade the datafiles in
place. The time it takes to upgrade in
this manner does not depend on the size of the database but rather the options
installed (partitioning, xml, ect.). If all pre-upgrade tasks are completed
before the actual time of the upgrade then a database upgrade using Database
Upgrade Assistant usually runs between 30 and 90 minutes. There are methods of moving
a database to 11g that can either increase or dramatically decrease this
downtime. Some of them involve running the DBUA or manually running upgrade
scripts and some do not. This paper highlights some of the different options
available. They can be modified and adjusted according to needs but the general
steps are outlined.
The following table summarizes the upgrade methods and rates
them (1 to5) in several areas. The difficulty in setup describes how hard
it would be to setup the scenario. The higher the difficulty means that there
is more that needs to be organized between teams, communicated, work done on both
the current operations environment (OE) and rehost operations environments
(ROE). The usability for testing
measures how difficult it would be to use the environment for testing
transactions multiple times. The higher the difficulty means that there is more
preliminary setup, steps to assure recoverability to designated point in times,
or steps to create a repeatable process. The higher the execution difficulty relates to the complexity the actual upgrade
process including tasks that need to be performed by non-dbas involved with
migrating to 11g on the day of cutover. The fallback ability describes how difficult it would be to fall back
to the OE 10g environment. As an example a 1 would indicate that you just need
to point the clients to the OE system. A 5 would indicate you would need to
recover from a previous backup. The final column estimated down time is just a SWAG (Scientific Wild Butt Guess) on
how long the database would be unusable. This can be adjusted as testing
occurs.
Method
|
Difficulty in Setup
|
Usability for Testing
|
Execution Difficulty
|
Fallback Ability
|
Estimated Down Time Guess
|
Datapump Import / Export
|
2
|
3
|
2
|
1
|
??? Depends on dataset size
|
Transportable Tablespaces
|
2
|
Not an Option
|
2
|
2
|
60 Minutes
|
Logical Standby Database
|
4
|
1
|
1
|
1
|
Less than 15 minutes
|
Upgrade using DBUA
|
2
|
Not an Option
|
3
|
5
|
Depends on options installed 30- 90
minutes
|
Upgrade using command line interface
|
2
|
Not an Option
|
5
|
5
|
Depends on options installed 30- 90
minutes
|
Physical Standby
|
4
|
3
|
2
|
1
|
Depends on options installed 30- 90
minutes
|
Oracle Streams / Golden Gate (not
really considered because of cost / setup)
|
>5
|
5
|
3
|
1
|
Less than 15 minutes
|
No comments:
Post a Comment