System Design Reference

Click a section to expand

Character Encoding & String Sizes ▶

Quick: English text ~1 byte/char (UTF-8). International ~2-3 bytes/char. Todo label ~50-100 bytes.

UTF-8 Encoding (Most Common)

Character Type	Bytes	Examples
ASCII	1	a-z, A-Z, 0-9, punctuation
Latin Extended	2	é, ñ, ü, ø
CJK	3	中, 日, 한
Emoji	4	😀, 🎉, 🚀

In-Memory String Representation

Different from disk! Languages decode to their own internal format.

Language	Internal Format	Bytes/Char
Python 3	Flexible (Latin-1/UCS-2/UCS-4)	1, 2, or 4
JavaScript	UTF-16	2 or 4
Java	UTF-16 (Latin-1 since Java 9)	1 or 2
Go / Rust	UTF-8	1-4

String Size Estimates

Content Type	Chars	Bytes (UTF-8)
Todo label	20-80	~50-100
Tweet	280	~280-560
Email subject	50-100	~50-150
Username	3-30	~3-30
URL	50-200	~50-200
Paragraph	~500	~500-800

Storage Units & Data Sizes ▶

Powers of 2

Power	Exact	Approx	Name
2^10	1,024	~1 thousand	1 KB
2^20	1,048,576	~1 million	1 MB
2^30	1,073,741,824	~1 billion	1 GB
2^40	1.1 trillion	~1 trillion	1 TB

Common Data Type Sizes

Type	Bytes	Range
boolean	1	true/false
int8 / byte	1	-128 to 127
int16 / short	2	-32K to 32K
int32 / int	4	-2.1B to 2.1B
int64 / long	8	±9.2 quintillion
float32	4	~7 decimal digits
float64 / double	8	~15 decimal digits
UUID	16	128-bit identifier

Latency Numbers ▶

Mental model: Memory = nanoseconds | SSD = microseconds | Network = milliseconds

Latency Reference

Operation	Time	Notes
L1 cache	~1 ns
L2 cache	~4 ns
L3 cache	~12 ns
RAM access	~100 ns	Main memory
SSD random read	~100 µs	NVMe: ~20 µs
HDD seek	~10 ms	Mechanical
Same datacenter RTT	~0.5 ms	Within rack: ~0.1 ms
Cross-region RTT	~50-150 ms	US East↔West: ~70 ms
Transatlantic RTT	~80-120 ms	US↔Europe

1 ms = 1,000 µs = 1,000,000 ns

Throughput & QPS ▶

Network Speeds

Connection	Speed
1 Gbps	125 MB/s
10 Gbps	1.25 GB/s
SSD write (SATA)	~500 MB/s
NVMe write	~3 GB/s

QPS Estimates

System	QPS	Notes
Single web server	1K-10K	Depends on complexity
Redis (single)	100K+	In-memory
MySQL/Postgres	5K-20K	Simple indexed queries

Time Conversions

Period	Seconds	1K QPS =
1 day	86,400	86M requests
1 month	~2.6M	2.6B requests
1 year	~31.5M	31.5B requests

Handy: 1 day ≈ 100K seconds, 1 year ≈ 30M seconds

Common Object Sizes ▶

User/Account Data

Field	Size
User ID	8-16 bytes
Username	~20 bytes
Email	~30 bytes
Password hash (bcrypt)	60 bytes
Basic user row	~200-500 bytes

Media Sizes

Type	Size
Avatar (100x100)	5-20 KB
Photo (web, 1080p)	100-500 KB
Photo (original)	2-10 MB
Video (1 min, 720p)	~50-100 MB
Video (1 min, 4K)	~300-500 MB
Audio (1 min, MP3)	~1 MB

Scaling Rules of Thumb ▶

Capacity Planning

Storage = users × data_per_user × retention_days

QPS_peak = QPS_avg × 2-3 (normal) or × 10 (viral)

Common Ratios

Metric	Typical
Read:Write	10:1 to 100:1
DAU:MAU	10-30%
Peak:Average traffic	2-3x
Cache hit rate (good)	95-99%

1 Million Users

If each user has...	Total
500 bytes (basic profile)	~500 MB
1 KB (profile + settings)	~1 GB
1 MB (with avatar)	~1 TB
10 MB (with photos)	~10 TB