Summary
Today I will show you how to identify the source problem to tune the query’s performance utilizing the unnecessary amount of IO, the UII indicator, and how the WHERE clause placed on a VIEW is evaluated.
A WHERE predicate on a view can be pushed into underlying operations. The historical case requires its actual plan and definitions to establish where filtering occurred; view syntax alone does not force reading every row first.
Performance tuning boosted the performance fivefold. The query is completed 4-5 times faster and utilizes 4-5 times fewer system resources:

1+ Billion rows INSERTED
Here is the target query for today, which populates a large fact table of the data mart.
– INSERT INTO … SELECT FROM …
– The query is run daily
– Over 1000M rows inserted (1 billion)
Poor performance detected
Query performance statistics stored in Teradata QueryLog signify that the query is suspicious and performs poorly in unnecessary IO, excessive Spool space, and CPU utilization.
Get the complete Teradata tuning book, free
Teradata Query Performance Tuning: The DWHPro Guide
By Roland Wenzlofsky & Artemios Vogiatzis
206 pages on statistics and demographics, access paths and indexes, query execution plans, and joins. Get the full PDF and practical Teradata updates from DWHPro.
Confirm your email to receive your download link. Privacy policy. Unsubscribe in any newsletter.
About the free bookTake a look at Teradata performance metrics:

What knowledge can we discover from this data?
What is particularly wrong, and how can we tune and improve the query?
A closer look at IO, source view DDL
First, let us describe what UII (Unnecessary IO indicator) is and what it can tell us.
UII = SumIO / (SumCPU * 1000) is a diagnostic ratio, not proof of unnecessary reads. Guard a zero CPU denominator and interpret it with row counts, access paths and the workload.
Let us examine source View DDL through which the target table is populated.
A-ha! Here we see that the latest six months are extracted from the view.
The tables are partitioned by columns used in WHERE clause:
But only last month’s data is inserted in the target table! The INSERT statement is as:
Here is the answer:
1. We might put the WHERE clause from the final query INSIDE the source VIEW DDL.
2. We might create an entirely new VIEW for this ETL process if some other users might not want any changes to the VIEW.
Improved query performance
This optimization gave us five times boost in performance. The query is completed 4-5 times faster and utilizes 4-5 times fewer system resources.
Detailed query metrics are below:

Artemiy Kozyr is Data Engineer at Sberbank, Moscow, with a Master’s Degree in CS.
He has five years of experience in Data Warehousing, ETL, and Visualization for Financial Institutions.
Contacts:
ar***********@***il.com
http://linkedin.com/in/artemiykozyr/
Validation scope: The view/table definitions and original workload are incomplete, so the fivefold benefit was not reproduced. The illustrated partition range ending in 2025 must be maintained for later data; handle out-of-range rows explicitly.
Need help with query performance or platform costs?
Diagnose plans, skew, spool pressure and expensive workloads. Agree a baseline and acceptance criteria before making changes.
Get the complete Teradata tuning book, free
Teradata Query Performance Tuning: The DWHPro Guide
By Roland Wenzlofsky & Artemios Vogiatzis
206 pages on statistics and demographics, access paths and indexes, query execution plans, and joins. Get the full PDF and practical Teradata updates from DWHPro.
Confirm your email to receive your download link. Privacy policy. Unsubscribe in any newsletter.
About the free book

Very helpful article, Thanks Artemiy.
Is there any similar article to deal with high PJI’s?