Vendors charge 60x more for hot storage than archive. The price gap reflects the economics of sub-millisecond instant access versus waiting hours for retrieval.
Defining storage as a commodity oversimplifies the issue. The spectrum is so wide, it’s like walking into a coffee shop and wanting a choice from night nurse to liquid caffeine rocket fuel.
A data-centric approach makes it easier to think about storage temperature. By matching performance to actual business requirements you ask “how quickly do you need this file” and “how often will you access it?”
Hot data serves data at millisecond latency. Warm storage gives instant access capacity at a 10x lower cost. Cold and Archive storage cuts the cost another 5x, but the time to first byte is inversely proportional to the latency.
The number one factor in multi-tier environments is that data can move between storage tiers without disrupting business operations. This is where we learn “the problem with your storage performance is your network”.
The real trap is thermal drift: files start hot and never cool down. Production assets from last quarter sit stale on premium storage. Nobody moves them because nobody measures the cost per file.
We need policies to automatically demote storage temperature based on access patterns. If nobody touches data for 31 days, it drops to warm. After 91 days, it goes cold. After a year, archive or delete.
The formula is simple: performance requirements multiplied by access frequency equals your tier. High performance times high frequency demands hot storage. Everything else is money you shouldn’t be spending.
What percentage of your storage budget pays for speed nobody uses?