The ×100 latency ladder

RAM ~100 ns, SSD ~100 µs, spinning disk ~10 ms — each rung roughly 100× slower.

Why it works

Five orders of magnitude between memory and disk is why caching dominates systems design — one disk seek costs ~100,000 RAM reads. The rungs are not a constant factor, though: memory to SSD is nearer 1000× than 100×, and the mnemonic understates the gap that matters most.

Do it exactly

Estimate with the rule, then check it against the calculator that models it properly.

Open Latency Ladder →

More in Computers

• 2¹⁰ ≈ 10³Every 10 bits multiplies by ~1,000 — kibi and kilo differ by just 2.4%.
• Each refresh rate step buys less than the lastFrame time is the reciprocal of refresh rate, so equal steps in Hz are not equal steps in latenc…
• Each subnet bit halves the hostsEvery bit moved from host to network halves the addresses — a /24 has 254 usable.
• RAID is not a backupRAID protects against disk failure — deletion, ransomware and fat fingers replicate instantly.

How slow is disk compared to RAM?

RAM ~100 ns, SSD ~100 µs, spinning disk ~10 ms — each rung roughly 100× slower. Five orders of magnitude between memory and disk is why caching dominates systems design — one disk seek costs ~100,000 RAM reads.

Browse all 176 rules of thumb →