Elapsed Time by Queries

 

Screen Shot 2020-09-09 at 9.48.16 AM

Why are your BI users are so frustrated?

- Double-click the points in this graph to copy the query-id for deeper analysis.


- This graph will allow you to detect and then remove queries that are outliers so you can save cluster resources.

- The queries above the dashed black line take more than 5 minutes.  Consider optimizing them.

- If you want to learn more about this graph click on the video for a detailed explanation.

 

ElapsedMemZeyde 1

CPU Utilization by Operator

Operators 2

 

Which Operators are costing you the most?

- Users can understand on how much resources are spent on scan, join, and aggregation.

- If the percentage of scan and filter is high these queries will benefit from indexing.

Input and Output Rows for Filtering Operators

Selectivity_Zeyde 3

 

What is selectivity and how does it impact performance & cost?

- Selectivity is defined by the ration between input and output rows.

- In this graph a dot represents a filter.  You will want to focus on everything that is below Y=X.  These are great candidates for indexing.

- Double-click the points in this graph to copy the query-id for deeper analysis.


- If you want to learn more about this graph click on the video for a detailed explanation.

Joins Distribution

Screen Shot 2020-09-09 at 9.49.07 AM

 

What's wrong with your Joins?

- Queries to the left of the black dashed line and above the orange dashed line should all use the REPLICATED join distribution type.


- Queries to the right of the orange dashed line perform joins with an incorrect table order. Ensure statistics are used, or rewrite the queries to flip the table sides, to boost performance and save cluster resources.


- Double-click the points in this graph to copy the query-id for deeper analysis.


- If you want to learn more about this graph click on the video for a detailed explanation.

Joins 4

 

Joins Selectivity

Joins 5

 

How selective are your Joins?

- Identify Join candidates for Dynamic Filtering.  A Varada contribution to the Presto open-source community.

- Double-click the points in this graph to copy the query-id for deeper analysis.


- If you want to learn more about this graph click on the video for a detailed explanation.

Don't have access yet? Try it free!

Download the Workload Analyzer

There's a New Standard for Data Virtualization

 

Meet Varada. With the use of dynamic analysis and adaptive indexing data architects are able to seamlessly accelerate and optimize workloads - resulting in optimal control over performance and cost.