Block storage and replication in Hadoop - Step-by-Step Execution

Concept Flow - Block storage and replication

File Input

↓

Split into Blocks

↓

Store Blocks on DataNodes

↓

Replicate Blocks

↓

Maintain Replication Factor

↓

Handle Node Failures

↓

Re-replicate Missing Blocks

↓

File Available for Access

A file is split into blocks, stored on different nodes, and each block is replicated to ensure reliability and availability.

Execution Sample

Hadoop

file = 'bigdata.txt'
blocks = split(file, block_size=128 * 1024 * 1024)  # 128MB in bytes
for block in blocks:
    store(block, datanode)
    replicate(block, replication_factor=3)

This code splits a file into blocks and stores each block on data nodes with replication.

Execution Table

Step	Action	Block	DataNode(s) Stored	Replication Count	Result
1	Split file into blocks	block_1	-	-	File split into block_1, block_2, block_3
2	Store block_1	block_1	DataNode1	1	block_1 stored on DataNode1
3	Replicate block_1	block_1	DataNode2, DataNode3	3	block_1 replicated to DataNode2 and DataNode3
4	Store block_2	block_2	DataNode2	1	block_2 stored on DataNode2
5	Replicate block_2	block_2	DataNode1, DataNode3	3	block_2 replicated to DataNode1 and DataNode3
6	Store block_3	block_3	DataNode3	1	block_3 stored on DataNode3
7	Replicate block_3	block_3	DataNode1, DataNode2	3	block_3 replicated to DataNode1 and DataNode2
8	DataNode2 fails	block_1, block_2, block_3	DataNode2 lost	Replication drops to 2 for affected blocks	Replication factor below 3 for blocks on DataNode2
9	Re-replicate missing blocks	block_1, block_2, block_3	New DataNode4	3	Blocks re-replicated to maintain replication factor
10	File available	-	-	-	All blocks have replication factor 3, file accessible

💡 All blocks have replication factor 3, ensuring data reliability and availability

Variable Tracker

Variable	Start	After Step 2	After Step 3	After Step 8	After Step 9	Final
block_1_replication_count	0	1	3	2	3	3
block_2_replication_count	0	1	3	2	3	3
block_3_replication_count	0	1	3	2	3	3
DataNodes_active	DataNode1, DataNode2, DataNode3	Same	Same	DataNode1, DataNode3	DataNode1, DataNode3, DataNode4	Same

Key Moments - 3 Insights

Why do we replicate blocks on multiple DataNodes?

Replication ensures that if one DataNode fails, the data is still available on other nodes, as shown in steps 3, 5, 7, and recovery in step 9.

What happens when a DataNode fails?

The replication count for blocks on that node drops (step 8), triggering re-replication to maintain the replication factor (step 9).

Why split a file into blocks instead of storing whole file on one node?

Splitting allows parallel storage and processing, improves fault tolerance, and fits large files across multiple nodes, as shown in step 1 and subsequent storage steps.

Visual Quiz - 3 Questions

Look at the execution table at step 3. What is the replication count of block_1 after replication?

A 1

B 2

C 3

D 0

💡 Hint

Check the 'Replication Count' column in row for step 3.

At which step does a DataNode failure occur causing replication to drop?

A Step 5

B Step 8

C Step 9

D Step 10

💡 Hint

Look for the step mentioning DataNode failure and replication drop.

If the replication factor was set to 2 instead of 3, how would the replication count change after step 3?

A Replication count would be 2

B Replication count would be 3

C Replication count would be 1

D Replication count would be 0

💡 Hint

Replication count matches the replication factor set during replication.

Concept Snapshot

Block storage splits large files into fixed-size blocks.
Each block is stored on multiple DataNodes.
Replication factor defines how many copies exist.
Replication ensures data availability and fault tolerance.
If a node fails, missing blocks are re-replicated.
This keeps data safe and accessible in Hadoop clusters.

Full Transcript

In Hadoop, large files are split into blocks for easier storage and processing. Each block is stored on a DataNode and replicated to other nodes to ensure reliability. The replication factor, often 3, means each block has three copies on different nodes. If a DataNode fails, the system detects the drop in replication and re-replicates the missing blocks to other nodes. This process keeps the file available and safe from data loss. The execution table shows steps from splitting the file, storing blocks, replicating them, handling node failure, and re-replication to maintain the replication factor.