Information Theory, Data Metrology & Binary vs Decimal Architecture
An authoritative guide to bits, bytes, octets, IEC binary prefixes (KiB, MiB, GiB), networking bandwidth, and OS storage discrepancies.
In digital computation and telecommunications, data quantification is grounded in Claude Shannon's foundational 1948 treatise on information theory. Because binary microprocessors address solid-state RAM and CPU registers using powers of two (2n), while storage manufacturers and networking engineers measure data using powers of ten (10n), accurate conversion is essential across computer science and cloud architecture.
1. The Discrepancy: Base-10 (SI) vs Base-2 (IEC Standard)
When a consumer purchases a "1 Terabyte" (1 TB) NVMe SSD, why does Windows report its capacity as approximately 931 GB?
- The Hardware Manufacturer (SI Standard): Solid-state drive manufacturers define 1 TB as exactly
10¹² = 1,000,000,000,000 bytes(Base-10). - The Operating System (Binary IEC Standard): The Windows kernel calculates file and partition sizes using binary powers of two:
1 TiB = 2⁴⁰ = 1,099,511,627,776 bytes.
Reported GiB = 1,000,000,000,000 / (1024 × 1024 × 1024) ≈ 931.32 GiB
2. Solved Step-by-Step Calculation Examples
Example A: Converting 1 Gigabyte (GB) to Megabytes (MB) and Kilobytes (KB)
-
1
Standard Base-10 Conversion:
1 GB = 1,000 MB = 1,000,000 KB. -
2
Binary Base-2 Conversion (GiB to MiB):
1 GiB = 1,024 MiB = 1,048,576 KiB.
Example B: Converting a 100 Mbps Internet Bandwidth to Real Download Time for a 5 GB File
-
1
Convert line speed from Megabits to Megabytes:
100 Mbps / 8 = 12.5 MB/s. -
2
Convert file size to Megabytes:
5 GB = 5,000 MB. -
3
Calculate transfer duration:
5,000 MB / 12.5 MB/s = 400 seconds = 6 minutes and 40 seconds(ignoring TCP/IP packet overhead).
3. Master Digital Storage Hierarchy Table
| Unit Name | Symbol | Bytes (B) - Base 10 | Bits (b) | IEC Binary Equivalent |
|---|---|---|---|---|
| Bit | b |
0.125 B (1/8) | 1 bit | 1 bit |
| Byte (Octet) | B |
1 B (Anchor) | 8 bits | 1 Byte |
| Kilobyte | KB |
1,000 B (10³) | 8,000 bits | 1 KiB = 1,024 B |
| Megabyte | MB |
1,000,000 B (10⁶) | 8,000,000 bits | 1 MiB = 1,048,576 B |
| Gigabyte | GB |
1,000,000,000 B (10⁹) | 8 × 10⁹ bits | 1 GiB = 1,073,741,824 B |
| Terabyte | TB |
10¹² B | 8 × 10¹² bits | 1 TiB = 1,099,511,627,776 B |
| Petabyte | PB |
10¹⁵ B | 8 × 10¹⁵ bits | 1 PiB = 1.125899 × 10¹⁵ B |
4. Real-World Cloud, Software & Telecom Applications
- Cloud Database Sizing & Billing: Cloud storage providers (Google Cloud Storage, AWS S3, Azure Blob) meter and invoice ingress, egress, and active tier storage in GiB-months and TiB-months, requiring exact byte-level conversions.
- Network Infrastructure & Video Streaming: 4K video bitrates require 15–25 Mbps bandwidth. Transmitting a 2-hour movie consumes approximately 13.5 to 22.5 Gigabytes of data transfer.
- Database Indexing & RAM Allocation: In-memory caches (Redis, Memcached) allocate memory structures in strict power-of-two page sizes (4 KiB, 2 MiB, 1 GiB huge pages).
5. Critical Digital Metrology Pitfalls
- Case Sensitivity (b vs B): Lowercase
bdenotes bits; uppercaseBdenotes bytes. Confusing 100 Mb with 100 MB results in an eight-fold (800%) error. - Ignoring Protocol Overhead: Raw payload transfer does not equal wire transmission. TCP/IP headers, TLS encryption framing, and packet retransmission add 3–5% overhead to real-world network throughput.