Get Started. It's Free
or sign up with your email address
Rocket clouds
SLOB by Mind Map: SLOB

1. Test Models

1.1. Physical Read Performance

1.1.1. Use a small db_cache_size

1.1.2. In 11.2.0.3 use specific parameters

1.1.2.1. alter system set "_db_block_prefetch_limit"=0

1.1.2.2. alter system set "_db_block_prefetch_quota"=0

1.1.2.3. alter system set "_db_file_noncontig_mblock_read_count"=0

1.1.2.4. alter system set compatible="11.2.0.2"

1.1.3. Should see db file sequential read as top wait event

1.2. Redo Generation Performance

1.2.1. Use large redo logs

1.2.2. Use a large db_cache_size

1.2.3. alter system set log_checkpoint_timeout=99999999

1.3. Physical Write Performance

1.3.1. Use small redo logs

1.3.2. Use a small db_cache_size

1.3.2.1. With small db_cache_size, don't need to include readers, will have physical reads

1.4. Logical I/O Performance

1.4.1. use a big db_cache_size (bigger than tables used)

1.4.2. avoid any auto memory management

2. Analysis Tools

2.1. awr_info.sh

2.1.1. Included in SLOB distribution

2.1.2. Summarizes multiple AWR reports

2.1.3. SLOB.R

2.1.3.1. Download

2.1.3.1.1. http://wp.me/p2Jp2m-4i

2.1.3.2. Interrogates awr_info.sh output in R

2.1.3.3. Requires R Statistical Environment

2.1.3.3.1. Download

3. Download

3.1. http://oaktable.net/articles/slob-silly-little-oracle-benchmark

4. General Setup

4.1. Eliminate other activity on database server

4.1.1. Disable automatic AWR snapshots

4.1.2. alter system set resource_manager_plan=''

4.2. Disable parallel unless testing parallel performance

4.2.1. alter system set parallel_max_servers=0

4.2.2. alter system set cpu_count=1

4.3. Switch logfiles (N*2+1) times before each run

4.3.1. N = number of redo logs

4.4. Do a throwaway SLOB run after a bounce to prime the database

4.5. Save awr.txt output after each run for further analysis

4.6. Scripts to automate multiple SLOB runs and save awr.txt

4.6.1. http://wp.me/p2Jp2m-29

4.7. Try misc/create_database_kit/ in SLOB to build a testing DB

4.8. Isolate UNDO blocks from table/index (chains) for write/REDO model by setting db_recycle_cache_size=N

5. Resources

5.1. Introducing SLOB: http://kevinclosson.wordpress.com/2012/02/06/introducing-slob-the-silly-little-oracle-benchmark/

5.2. Collected links/tips/tools: http://wp.me/p2Jp2m-3n

5.3. Yury Velikanov's SLOB index: http://www.pythian.com/blog/my-slob-io-testing-index/

5.4. SLOB LinkedIn Group: http://t.co/6O4ynhjB

5.5. Karl Arao's TiddlyWiki: http://karlarao.tiddlyspot.com/#[[cpu%20-%20SillyLittleBenchmark%20-%20SLOB]]

6. Uses

6.1. Benchmark one storage unit vs another

6.2. Compare performance of various filesystems

6.2.1. Don't mix LUNs; use same LUN for each FS

6.3. Identify max throughput possible from your setup

7. Non-SLOB tools

7.1. hammerora

7.2. ORION

7.3. swingbench

8. Other Considerations

8.1. Distributing your SLOB results/AWRs may be a legal issue with Oracle Corp

8.1.1. Scale data against a baseline?