|
|
back to the main page
Oracle Database 11g: Rich Niemiec’s Favorite New Features
In 1987 Rich Niemiec came to Oracle to help build client-server systems on Oracle Database Version 6. He went on to become one of the world’s leading advocates of Oracle technology. He has written several Oracle-related best-sellers and recently released an update of his definitive guide to Oracle performance tuning, Oracle Database 10g Performance Tuning Tips & Techniques. He is the former president of the International Oracle Users Group (IOUG) and current CEO of TUSC, an Oracle expert-level consulting firm that has been on Inc. magazine’s list of the 500 fastest-growing privately owned companies. As Oracle Database 11g is about to launch, we asked Niemiec to talk about its best features.
|
|
Insider:
|
Tell us about planned features, big or small, that will have an immediate impact in the data center.
|
|
Niemiec:
|
First, I want to give my overall impression of Oracle Database 11g. I’m excited about Oracle Database 11g, because I think it manages the business systems of the future. The amount of data continues to grow while the pace of change accelerates. We’ll need systems that give us a way to visualize and manage huge data loads. We’ll need systems that help us meet that change with confidence. We’ll need systems that have the intelligence to manage themselves to some degree. The goal is for Oracle Database 11g to do all these things.
The first planned feature I would mention—and this might be the best one—is workload capture and replay. People are constantly making changes in their data centers: implementing migrations and upgrades or changing hardware, operating system, and applications. They need a way to ensure consistent or better behavior when they make these changes. We expect Oracle Database 11g’s workload capture-and-replay feature to be huge.
|
|
Insider:
|
Tell us more about what workload capture and replay does.
|
|
Niemiec:
|
Workload capture and replay is a simple, menu-driven way of capturing a system’s workload over time and then replaying it in precisely the same way in which it was captured. You can capture workload for a minute, an hour, or several days. Then, when you make changes in your system, you can test your real day-to-day workloads against the new system. You can make changes with confidence.
This is also great for tuning the database. For example, you might say, “Our system runs great, but from 12 o’clock to 1 o’clock, it slows down, and we’re not sure why.”
So from 12 o’clock to 1 o’clock, you could run the capture-and-replay feature and build a tuning set for your exact workload for that period of time. Then, when you make a change, you’ll know how it will affect each of the various parameters in your system. The system would tell you, among other things, the number of block reads it’s going to change or how the CPU is going to change and would graph it out to show you the before and after for this whole group of SQL statements.
So this planned feature is going to be great for both managing change and tuning existing systems.
|
|
Insider:
|
Any other planned tuning features?
|
|
Niemiec:
|
One of my favorite tuning features planned for Oracle Database 11g is something called the invisible index. I’ll be able to create or alter an index to be visible or invisible. Indexes are usually visible to everyone, but now, let’s say I see that somebody has overindexed something, which is often the case with third-party applications. Let’s say it’s a single table that you know doesn’t need 50 indexes, but you’re afraid to drop any. So you could say, “Oh, these 10 indexes—I’m pretty sure they’re worthless.” And instead of dropping them, you just make them invisible.
Then, of course, as the table is updated, the index continues to be updated too. So, if you eventually say, “Oh, that was a bad idea” and you want to get the indexes back, you don’t have to rebuild them—you just make them visible. This is a phenomenal feature.
|
|
Insider:
|
Tell us about features in Oracle Database 11g that Oracle expects will help make customers more productive.
|
|
Niemiec:
|
SQL Repair Advisor is one of Oracle Database 11g’s many self-managing features. If you come across an error in your code, SQL Repair Advisor will try to find a route around it. It automatically reparses the statement and tries a different path, similar to the workaround you’d get if you called Oracle Support. This is great for small and medium businesses that don’t have a lot of time to spend on managing their system.
Another time-saver is an enhancement called “virtual columns,” which you would use if you were always doing calculations in a column (such as salary x 100, or time x 24 hours). You can now have a column with the calculation built in. So if you’re running a large batch process, for example, and you’re doing the same calculation over and over, you can store the calculation instead of continually recalculating it. It’s another way Oracle Database 11g helps customers get faster results.
|
|
Insider:
|
Do some of these manageability enhancements affect grid computing?
|
|
Niemiec:
|
There is a planned enhancement to the Automatic Database Diagnostic Monitor (ADDM) that’s going to help with managing grids. ADDM, which is pronounced “Adam,” automatically analyzes the performance of the database at regular intervals to detect and diagnose common performance problems, such as CPU bottlenecks or connection management issues.
In Oracle Database 11g, the optimizer continues to get better. For example, instead of taking statistics for just one column, the optimizer now takes statistics for multiple columns that relate to each other. So, for example, if I’m in retail, a season change also signals an inventory change from, say, snowshoes to swimsuits. Now, instead of looking at the season column and the inventory columns separately, the optimizer can join those two columns into multicolumn statistics to facilitate much smarter and faster decisions.
|
|
Insider:
|
How does Oracle Database 11g boost performance?
|
|
Niemiec:
|
I’ll illustrate with one core component of Oracle Database called the optimizer—one of the elements of Oracle Database that makes it great. Every time you search for data, the optimizer asks, “How should I access that data to make it fast?” With every search, it learns and fine-tunes the fastest way to retrieve that data. It’s the most complex part of Oracle Database, and it’s probably 10 to 15 years ahead of any other database program.
In Oracle Database 11g, the optimizer continues to get better. For example, instead of taking statistics for just one column, the optimizer now takes statistics for multiple columns that relate to each other. So, for example, if I’m in retail, a season change also signals an inventory change from, say, snowshoes to swimsuits. Now, instead of looking at the season column and the inventory columns separately, the optimizer can join those two columns into multicolumn statistics to facilitate much smarter and faster decisions.
I’ve studied the rest of the market, and I give talks about where Oracle is, compared to its competition. When I looked at Oracle Database 10g, it was already more than a decade ahead of its closest competitor. With Oracle Database 11g, I think it adds another five years.
|
|
This article is for informational purposes only and may not be incorporated into a contract or agreement. It does not represent a commitment to deliver any code or functionality and should not be relied on in making a purchase decision. The development, release, and timing of Oracle’s products remain at the sole discretion of Oracle.
|
|
|
|
|
