NOTE
2.3 Persistent Storage vs. Memory
A qualitative comparison of DRAM, SSDs, and HDDs across latency, throughput, capacity, cost, persistence, and sequential vs. random access.
This is a historical learning note and may contain outdated or incomplete understanding.
1. Memory and Persistent Storage Solve Different Problems
DRAM is fast volatile working memory. SSDs and HDDs provide persistent storage with much larger capacity at lower cost per byte.
A useful qualitative comparison is:
| Property | DRAM | SSD | HDD |
|---|---|---|---|
| persistence | volatile | persistent | persistent |
| latency | lowest | higher | highest |
| random access | very fast | good | expensive due to mechanics |
| sequential throughput | high | high | much better than HDD random I/O |
| capacity/cost per byte | expensive | medium | cheapest at large capacity |
2. Sequential vs. Random Access
Access pattern matters greatly, especially for storage.
HDDs pay seek and rotational costs, so random I/O can be orders of magnitude slower than long sequential transfers.
SSDs remove mechanical seek time, but random access still has controller, flash-translation, queueing, write-amplification, and block-management costs.
DRAM also benefits from locality and row-buffer/cache behavior; “random vs. sequential” is not irrelevant to memory.
3. Do Not Use One Fixed Speed Ordering
Statements such as “random DRAM equals sequential disk” are not durable hardware facts.
Actual latency and throughput depend on:
- hardware generation;
- queue depth;
- transfer size;
- cache/page-cache hits;
- filesystem and device scheduler;
- NUMA placement;
- access pattern.
Measure the actual workload instead of relying on a single universal ratio.