How Big Query Stores Data and Executes a Query (In Simple Words)….

In this blog, we’ll break down how Big Query stores data and how it executes a query, using simple language and real-world analogies.

How Big Query Stores Data:-

Big Query uses columnar storage, not row-based storage like traditional databases (MySQL, PostgreSQL).

🔹 Row-based storage which used for traditional database is good for transaction but bad for analytics.

🔹 Columnar storage is used to store data column by columns this is more powerful than row-based storage because.

• Queries usually read only a few columns
• It scans only required columns
• Less data read = faster queries + lower cost

Since columns store similar data numbers with numbers, strings with strings for which Big Query compresses data very efficiently , As result.

• Less storage used
• Less data scanned
• Faster performance

You don’t need to manage compression — Big Query does it for you.

Big Query doesn’t store your data on one machine it used distributed storage method.

• Data is split into multiple small parts
• Stored across many machines

Example:-

A huge book broken into pages and stored in many libraries — you can read multiple pages at the same time.

How Big Query Executes a Query:-

Step 1: Query Validation

Big Query first:

• Checks SQL syntax
• Validates tables and columns
• Estimates how much data will be scanned (shown before execution)

If there’s an error, execution stops here and the error details is going to shown in Big Query console.

Step 2: Query Planning

• Breaks query into stages
• Decides join strategies
• Determines parallel execution

Step 3: Distributed Execution

This is where Big Query .

• Query is split into many small tasks
• Tasks run in parallel
• Each worker processes a chunk of data

This is powered by:

• Google’s Dremel engine

Step 4: Slot Allocation

Big Query uses slots (units of compute power).

• More slots = faster execution
• Fewer slots = longer execution

Depending on pricing:

• On-demand: slots allocated automatically
• Flat-rate: you reserve slots

Step 5: Result Delivery

• Results are returned to UI / API
• we can written to a destination table
• we can exported to Cloud Storage

Why Big Query Is So Fast:-

Big Query is fast because it combines:

✅ Columnar storage ✅ Automatic compression ✅ Distributed storage ✅ Massive parallel execution ✅ Serverless architecture

You don’t manage:

• Servers
• Indexes
• Storage tuning

You just write SQL.

Key Takeaways

• BigQuery stores data column-wise, not row-wise
• It scans only the columns you need
• Data is compressed and distributed automatically
• Queries run in massive parallel
• Cost and performance depend on data scanned