Oracle surveyed our installed base and we found that regardless of company size or number of databases managed, the tasks that a DBA needs to perform fall into one of three main categories.
Delivering sustained, optimal performance and uptime is critical to the business users of your databases.
Even while performance management is going on, your environment itself is changing. You may be adding indexes to improve the performance of a particular report, or you may be changing from large Unix servers to a grid of Linux servers or changing out your Storage Area Network or just applying routine patches to your database or your operating system. Amidst all this change, the database and business applications must continue to perform and be available to your business users. It is the role of the DBA to manage all this change in an orderly fashion.
And while this is all happening, you still have the manage the ongoing tasks in your environment like space management, user management, and the provisioning and testing of new systems.
These tasks consume nearly all of the time for DBAs leaving little time for training or professional development. And the number of databases being managed generally only goes one way – up. All of these demands make it necessary for intelligent tools to assist DBAs and that is what Oracle is here to talk with you about today.
Database automation and manageability starts inside the database and starting in the 9i release, Oracle has added instrumentation to the Oracle database to gather critical statistics on usage and workload. In 10g, Oracle has added a number of advisors which make recommendations to help resolve issues and increase manageability. There are also a number of auto-tuning features in 10g to make tuning simpler and faster.
Now, in 11g, there are even more auto tuning features for SQL tuning and schema and RAC and recoverability and replication.
Oracle has been making steady progress in increasing the manageability of the Oracle Database. This graph compares the amount of time and the number of steps required for typical operations in Oracle in the 9i, 10g, and 11g releases. 11g now take 26 percent less time and 31 percent fewer steps that its predecessor, 10g.
This is yet another reason to upgrade to the latest version of the database supported by your application.
Now, let’s look at some of the top challenges DBAs face in the area of performance management and lets break these down into to Performance Diagnostics and Application Tuning.
ADDM is a core part of the diagnostic infrastructure in the Oracle Database and it was introduced in Oracle 10g and it has further evolved in Oracle Database 11g. ADDM starts off with analysis of snapshots of key workload metrics taken in the Database. These snapshots include critical performance information pertaining to the database kernel and the database workload as well as at the operating system level. ADDM runs on a regular basis (by default once an hour) and it analyzes this information and makes recommendations on problems that it identifies.
For example, when there are SQL load issues, the SQL Advisors make recommendations on how to tune the SQL statements. Also when there are IO or CPU issues, ADDM gives advice on system resource optimization. And if you are running a RAC cluster, ADDM analyzes your complete RAC infrastructure including your interconnect and it offers advice there as well.
ADDM is part of the database kernel, so it is always up to date an in synch with your version of the database.
ADDM uses a Classification Tree methodology and it brings to your attention only those things which require your attention.
So, let’s see how a DBA 1.0 solution compares to a DBA 2.0 solution in a particular scenario. Lets say that one day you begin to suspect that you have a hard parsing problem with your database due to some recent application usage changes. There is a very detailed methodology to follow if you want to identify hard parsing and to verify that it is placing a substantial load on your database.
The DBA 2.0 approach would be to check the Database Control page for any recommendations from ADDM. If their were an excessive number of hard parses happening, ADDM would already know and you would see a recommendation to use cursor_sharing.
A task that DBAs have to perform all the time is patching. Oracle Enterprise Manager has the ability to connect to Oracle Metalink and analyze and download the patches you need, apply them in your test environment, promoted them to your staging area, and deploy them into production, and verify that they were applied correctly. This patch lifecycle applies to single patches, patch sets, and Critical Patch Updates.
Also, with Oracle Real Application Clusters, certain patches can be applied in a rolling fashion to each node in the cluster to eliminate downtime.
And with Oracle Enterprise Linux, the ability to download and apply O/S patches is totally automated from beginning to end requiring no intervention at the console.
Traditionally, managing deployments of new systems has been a very manual process which required a lot of time from the DBA. Oracle has embraced the notion of gold image-based provisioning. Gold images help simplify the creation and deployment of new systems from known, certified configurations which are created by the DBAs and stored in the configuration repository.
Single click RAC deployment is supported for adding or removing capacity from your RAC cluster as you need it. This covers the database as well as the clusterware. It configures the entire stack needed for your RAC cluster.
Another key requirement for enterprises is to be able to securely deploy test systems with sensitive production data masked.
This enables the business to clone copies of the production database for testing or training, but we business critical information masked so that it cannot be understood.
The process of masking has no impact on the production database since it happens on the staging instance.
There are masking templates for common data types like SSN or Credit Card number.
And referential integrity is maintained. If related columns need to be maskes, they will be masked identically so that RI is intact.
Routine Maintenance falls into four key categories.
1). Fault Diagnostics are used to identify failures or errors and to get those issues identified and resolved in a timely manner.
2). Resource Management is the allocation of resources like memory and disk as well as logical operations like space mgmt.
3). SQL Optimization deals with statistics gathering and versioning and SQL tuning and plan management
4). Backup and Recovery deals with maintain a highly maintainable and recoverable system.
If time permits, discuss a few of the features from each of the four areas.
© 2010, www.oracledatabase12g.com. 版权所有.文章允许转载,但必须以链接方式注明源地址,否则追究法律责任.
相关文章 | Related posts:
- Oracle Database 11g: Server Manageability Student Guide
- EXTREME PERFORMANCE:Exadata & Database Machine
- Oracle Database 10G – Manageability
- Extreme Performance with Oracle Data Warehousing
- Extreme Intelligence:From Storage to Scorecard
- OCP-8i DBA:Oracle Architectural Components
- OCP-8i DBA:Getting Started with the Oracle Server
最新评论