GROUP and JOIN operations in Hadoop - Step-by-Step Execution

Concept Flow - GROUP and JOIN operations

Input Data

↓

Map Phase

↓

Shuffle & Sort

↓

Group by Key

↓

Join Keys

→Reduce Phase for JOIN

↓

Output Data

Data flows from input through map, then shuffle groups data by keys, then reduce performs GROUP aggregation or JOIN combining.

Execution Sample

Hadoop

map(key, value):
  emit(key, value)

reduce(key, values):
  if GROUP:
    emit(key, aggregate(values))
  if JOIN:
    emit(key, combine(values_from_both_datasets))

Map emits key-value pairs; reduce aggregates values for GROUP or combines values for JOIN by key.

Execution Table

Step	Phase	Input	Action	Output
1	Map	Dataset1: (A,1), (B,2)	Emit key-value pairs	(A,1), (B,2)
2	Map	Dataset2: (A,3), (C,4)	Emit key-value pairs	(A,3), (C,4)
3	Shuffle & Sort	(A,1), (B,2), (A,3), (C,4)	Group by key	A: [1,3], B: [2], C: [4]
4	Reduce (GROUP)	A: [1,3]	Sum values	(A,4)
5	Reduce (GROUP)	B: [2]	Sum values	(B,2)
6	Reduce (GROUP)	C: [4]	Sum values	(C,4)
7	Reduce (JOIN)	A: values from both datasets	Combine values	(A,(1,3))
8	Reduce (JOIN)	B: values only from Dataset1	No match in Dataset2	(B,(2,null))
9	Reduce (JOIN)	C: values only from Dataset2	No match in Dataset1	(C,(null,4))
10	End	All keys processed	Output final grouped or joined data	GROUP output: (A,4), (B,2), (C,4) JOIN output: (A,(1,3)), (B,(2,null)), (C,(null,4))

💡 All keys processed, reduce phase completes GROUP aggregation or JOIN combination.

Variable Tracker

Variable	Start	After Map 1	After Map 2	After Shuffle	After Reduce GROUP	After Reduce JOIN
Dataset1	[(A,1), (B,2)]	[(A,1), (B,2)]	[(A,1), (B,2)]	Grouped by key	Aggregated sums	Used for join
Dataset2	[(A,3), (C,4)]	[(A,3), (C,4)]	[(A,3), (C,4)]	Grouped by key	Aggregated sums	Used for join
Grouped Data	{}	{}	{}	{A:[1,3], B:[2], C:[4]}	{A:4, B:2, C:4}	{A:(1,3), B:(2,null), C:(null,4)}

Key Moments - 3 Insights

Why do keys appear multiple times before the reduce phase?

Because the map phase emits key-value pairs for each record, keys repeat until shuffle groups them together as shown in execution_table step 3.

How does the reduce phase know whether to do GROUP or JOIN?

The reduce function logic distinguishes GROUP by aggregating values or JOIN by combining values from multiple datasets, as seen in steps 4-9.

What happens if a key exists in only one dataset during JOIN?

The reduce phase outputs the key with null for missing dataset values, shown in steps 8 and 9.

Visual Quiz - 3 Questions

Look at the execution table at step 3. What is the grouped output for key 'A'?

A [2]

B [1,3]

C [4]

D [1,2,3]

💡 Hint

Check the 'Shuffle & Sort' phase output in execution_table row 3.

At which step does the reduce phase output the sum for key 'B' in GROUP operation?

A Step 7

B Step 3

C Step 5

D Step 9

💡 Hint

Look for 'Reduce (GROUP)' phase output for key 'B' in execution_table.

If Dataset2 had no key 'C', how would the JOIN output for 'C' change?

A No output for 'C'

B (C,(null,4))

C (C,(null,null))

D (C,(null,null)) with empty values

💡 Hint

Refer to execution_table steps 8 and 9 for keys missing in one dataset.

Concept Snapshot

GROUP and JOIN in Hadoop:
- Map emits key-value pairs
- Shuffle groups data by keys
- Reduce aggregates for GROUP (e.g., sum)
- Reduce combines datasets for JOIN
- Keys missing in one dataset get nulls in JOIN
- Output is grouped or joined data by key

Full Transcript

This visual execution shows how GROUP and JOIN operations work in Hadoop. First, the map phase emits key-value pairs from input datasets. Then, shuffle and sort group these pairs by key. In the reduce phase, for GROUP operations, values for each key are aggregated, such as summing numbers. For JOIN operations, values from both datasets are combined by key, with nulls used when a key is missing in one dataset. The execution table traces each step, showing inputs, actions, and outputs. Variable tracking shows how data changes from start to finish. Key moments clarify common confusions about repeated keys, reduce logic, and handling missing keys. The visual quiz tests understanding of grouping, reduce outputs, and join behavior with missing keys. This step-by-step trace helps beginners see exactly how Hadoop processes GROUP and JOIN operations.

Practice

1. What does the GROUP BY operation do in Hadoop data processing?

easy

A. It organizes data by keys to summarize or analyze groups.

B. It merges two datasets based on matching keys.

C. It deletes duplicate records from a dataset.

D. It sorts data in ascending order.

Solution

1. Step 1: Understand the purpose of GROUP BY

   The GROUP BY operation groups rows that have the same values in specified columns into summary rows.

2. Step 2: Differentiate from JOIN

   JOIN merges datasets based on keys, but GROUP BY organizes data within one dataset by keys.

3. Final Answer:

   It organizes data by keys to summarize or analyze groups. -> Option A

4. Quick Check:

   GROUP BY = Organize by keys [OK]

Hint: GROUP BY groups data by keys for summaries [OK]

Common Mistakes:

• Confusing GROUP BY with JOIN
• Thinking GROUP BY deletes duplicates
• Assuming GROUP BY sorts data

2. Which of the following is the correct syntax to join two datasets table1 and table2 on the key id in HiveQL (Hadoop SQL)?

easy

A. SELECT * FROM table1 GROUP BY table2.id;

B. SELECT * FROM table1 JOIN table2 ON table1.id = table2.id;

C. SELECT * FROM table1 WHERE table1.id = table2.id;

D. SELECT * FROM table1, table2 ON table1.id = table2.id;

Solution

1. Step 1: Identify correct JOIN syntax

   The standard JOIN syntax in HiveQL uses ON to specify the join condition: JOIN table2 ON table1.id = table2.id.

2. Step 2: Eliminate incorrect options

   SELECT * FROM table1 WHERE table1.id = table2.id; uses WHERE incorrectly for join; SELECT * FROM table1 GROUP BY table2.id; uses GROUP BY wrongly; SELECT * FROM table1, table2 ON table1.id = table2.id; incorrectly uses ON with comma-separated tables.

3. Final Answer:

   SELECT * FROM table1 JOIN table2 ON table1.id = table2.id; -> Option B

4. Quick Check:

   JOIN with ON condition = SELECT * FROM table1 JOIN table2 ON table1.id = table2.id; [OK]

Hint: Use JOIN ... ON for correct Hadoop SQL joins [OK]

Common Mistakes:

• Using WHERE instead of ON for JOIN
• Confusing GROUP BY with JOIN
• Using ON clause with old-style comma join

3. Given two datasets in Hive:
table1(id INT, value INT)
table2(id INT, category STRING)
What is the output of this query?

SELECT t2.category, SUM(t1.value) FROM table1 t1 JOIN table2 t2 ON t1.id = t2.id GROUP BY t2.category;

medium

A. Sum of values grouped by id from table1.

B. List of categories with no aggregation.

C. Sum of values grouped by category from table2.

D. Error due to missing GROUP BY column.

Solution

1. Step 1: Understand JOIN and GROUP BY

   The JOIN merges rows from table1 and table2 where ids match. Then GROUP BY groups results by category.

2. Step 2: Analyze aggregation

   SUM(t1.value) calculates total value per category from joined data.

3. Final Answer:

   Sum of values grouped by category from table2. -> Option C

4. Quick Check:

   JOIN + GROUP BY category = sum per category [OK]

Hint: JOIN then GROUP BY aggregates data by joined keys [OK]

Common Mistakes:

• Ignoring GROUP BY column in SELECT
• Confusing grouping by id vs category
• Expecting list without aggregation

4. You wrote this Hive query:

SELECT t1.id, COUNT(*) FROM table1 t1 JOIN table2 t2 ON t1.id = t2.id;

But it gives an error. What is the likely cause?

medium

A. Missing GROUP BY clause for aggregation.

B. Incorrect JOIN syntax.

C. COUNT(*) cannot be used with JOIN.

D. Table aliases are not allowed.

Solution

1. Step 1: Identify aggregation usage

   COUNT(*) is an aggregate function and requires GROUP BY for non-aggregated columns in SELECT.

2. Step 2: Check query structure

   Since t1.id is selected without aggregation, GROUP BY t1.id is needed.

3. Final Answer:

   Missing GROUP BY clause for aggregation. -> Option A

4. Quick Check:

   Aggregate needs GROUP BY [OK]

Hint: Use GROUP BY when selecting non-aggregated columns with COUNT [OK]

Common Mistakes:

• Forgetting GROUP BY with aggregate functions
• Thinking JOIN syntax is wrong
• Assuming COUNT(*) disallowed with JOIN

5. You have two large datasets:
sales(order_id, product_id, amount)
products(product_id, category)
You want to find total sales amount per category but only for categories with total sales above 1000. Which Hive query correctly achieves this?

hard

A. SELECT p.category, SUM(s.amount) FROM sales s JOIN products p ON s.product_id = p.product_id HAVING SUM(s.amount) > 1000;

B. SELECT p.category, SUM(s.amount) FROM sales s JOIN products p ON s.product_id = p.product_id WHERE SUM(s.amount) > 1000 GROUP BY p.category;

C. SELECT p.category, SUM(s.amount) FROM sales s JOIN products p ON s.product_id = p.product_id GROUP BY p.category WHERE SUM(s.amount) > 1000;

D. SELECT p.category, SUM(s.amount) FROM sales s JOIN products p ON s.product_id = p.product_id GROUP BY p.category HAVING SUM(s.amount) > 1000;

Solution

1. Step 1: Join sales and products on product_id

   This merges sales with product categories.

2. Step 2: Group by category and sum amounts

   GROUP BY p.category with SUM(s.amount) calculates total sales per category.

3. Step 3: Filter groups with HAVING

   HAVING filters groups after aggregation; WHERE cannot filter aggregated sums.

4. Final Answer:

   SELECT p.category, SUM(s.amount) FROM sales s JOIN products p ON s.product_id = p.product_id GROUP BY p.category HAVING SUM(s.amount) > 1000; -> Option D

5. Quick Check:

   Use HAVING to filter aggregated groups [OK]

Hint: Use HAVING to filter after GROUP BY aggregation [OK]

Common Mistakes:

• Using WHERE instead of HAVING for aggregation filter
• Omitting GROUP BY when aggregating
• Incorrect JOIN condition