[001] storage

How much disk does one day of access logs use?

easy loggingdiskretention

The scenario

You operate a service that handles roughly 200 million requests per day. Every request emits one line to an access log (think nginx combined format: IP, timestamp, method, path, status code, user-agent, response time).

How much disk does one day of raw (uncompressed) access logs consume?

hints

your answer

$answer:GB/day

$ sre-cheatsheet

# latency numbers

L1 cache ref	0.5 ns
Branch mispredict	5 ns
L2 cache ref	7 ns
Mutex lock/unlock	25 ns
Main memory ref	100 ns
Compress 1 KB	10 µs
Send 2 KB / 1 Gbps	20 µs
SSD random read	150 µs
Read 1 MB / memory	250 µs
Same-DC round trip	500 µs
Read 1 MB / SSD	1 ms
Disk seek	10 ms
Read 1 MB / network	10 ms
Read 1 MB / disk	30 ms
CA → EU → CA	150 ms

# powers of two

2¹⁰	1 K (~thousand)
2²⁰	1 MB (~million)
2³⁰	1 GB (~billion)
2³²	4 GB (max uint32)
2⁴⁰	1 TB (~trillion)
2⁶⁴	~18 EB

# availability

SLA	downtime / year	/ month
99%	3.65 days	7.3 h
99.9%	8.7 hours	43 min
99.99%	52 min	4.4 min
99.999%	5.3 min	26 s

# rules of thumb

Writes are ~40× more expensive than reads
Compression saves ~2× bandwidth
Memory >> SSD >> Network >> Disk
Prefer parallel over serial I/O — up to 10× faster
A server handles ~10k req/s for simple endpoints
Typical DB query: 1–10 ms
HTTP overhead: ~1 ms on same continent