Scribd
Upload
6 views4 pages

SQL

Uploaded by

Mofizur Rahman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
6 views4 pages

SQL

Uploaded by

Mofizur Rahman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Yes, the Excel sheet you provided is an excellent starting point for

identifying high-impact SQL queries. Here's a structured approach to


analyze and prioritize queries effectively:

Step 1: Key Metrics to Focus On

Primary Metric: Elapsed Time

 Why?: Elapsed Time represents the total duration the query took,
encompassing CPU time, I/O waits, and other overheads.

 Approach:

o Focus on queries with the highest Elapsed Time to address


the biggest bottlenecks.

Secondary Metric: Elapsed Time per Execution

 Why?: High Elapsed Time/Exec indicates that each execution of the


query is expensive.

 Approach:

o Prioritize queries with high Elapsed Time/Exec, as they likely


represent inefficient SQL plans.

Tertiary Metric: Buffer Gets

 Why?: Buffer Gets measures logical I/O, which shows how many
database blocks the query accessed.

 Approach:

o High Buffer Gets often correlate with inefficient execution


plans, such as table scans or poorly filtered joins.

Additional Metrics to Consider

 Executions: A query with moderate Elapsed Time but millions of


executions may still be a critical bottleneck.

 CPU Time: Helps identify CPU-bound queries.

 Direct Writes/User I/O Waits: Indicates heavy disk activity and


potential I/O bottlenecks.

Step 2: Prioritization

High Impact Queries


Sort the Excel sheet by Elapsed Time (descending) to identify the most
time-consuming queries.

Balance with Frequency

Look at Elapsed Time/Exec alongside Executions:

 High Elapsed Time + High Executions = Major Bottleneck (likely


application-driven inefficiencies).

 High Elapsed Time/Exec + Low Executions = Likely a poorly


optimized heavy query.

Step 3: Analyze Execution Plans

For each SQL ID identified from Step 2:

1. Retrieve Execution Plans:

2. SELECT *

3. FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR('<SQL_ID>', NULL,


'ALLSTATS LAST'));

4. Key Elements to Analyze:

o Full Table Scans: Look for TABLE ACCESS FULL.

o Index Usage: Confirm if INDEX RANGE SCAN or INDEX


UNIQUE SCAN is used.

o Joins: Check join types (nested loops vs. hash joins).

o Sorts and Aggregations: Identify costly operations like


SORT ORDER BY.

5. Optimization Recommendations:

o Add/adjust indexes for filter and join columns.

o Rewrite queries for efficiency (e.g., replacing DISTINCT with


analytic functions).

Step 4: Iterative Improvement

1. Start with Top Queries:

o Focus on the top 5–10 queries with the highest combined


impact (Elapsed Time + Buffer Gets).
2. Validate Improvements:

o Test changes (e.g., index creation, query rewrites) in a staging


environment.

o Re-run execution plans to ensure the query performance


improves.

3. Monitor in Production:

o Use AWR or Active Session History (ASH) to validate post-


change performance.

Sample Prioritization Table

Elapsed
Elapsed Executi Priori
SQL ID Time/Ex Buffer Gets Notes
Time (s) ons ty
ec (ms)

Very
9rttw1358g 262,576,5 773,363. 143,311,119 frequent
339,524 High
ywj 98.6 9 ,546 execution,
needs index

Heavy I/O;
g4x3b5vnb 564,938,9 5,539,92 117,011,286 optimize
101,976 High
5zf0 44.5 0.6 ,818 table
partitions

Costly joins;
15stf55d7m 179,688,9 64,088,840, 10,754,7 Mediu consider
16,707.8
pxr 66.4 690 86 m denormaliza
tion

Large table
2a5dnp0sr7 726,832,6 244,683. 38,746,473, 2,970,50 scans;
High
7h3 57.8 1 111 6 partition
required

Presentation Plan

Focus Areas

1. Current Performance Issues:


o Highlight the top resource-consuming queries and their
metrics.

2. Root Causes:

o Lack of indexes, high buffer gets, frequent executions.

3. Proposed Solutions:

o Indexes, query rewrites, table partitioning.

4. Expected Outcomes:

o Reduction in Elapsed Time and Buffer Gets, lower CPU and I/O
waits.

Would you like me to help prioritize specific SQL IDs or refine the analysis
further?

You might also like