Walk into a 1960s data center and you’d see two stars of the show:
- Big, swappable disk packs that could pass for props in a sci-fi set.
- Rows of reel-to-reel tape drives whirring away like a NASA control room.
Together, they defined the decade: disks for speed, tape for scale. And that mix birthed an idea that still drives storage strategy today — tiering.
Performance & Economics (Then → Now)
Removable disk packs:
- Capacity: 5–10 MB per pack.
- Access time: tens to low hundreds of milliseconds — fast for the era.
- Cost: even with falling prices, disks translated to $100,000+ per GB in today’s dollars when factoring in media, drives, and service.
- Ops reality: delicate to handle, required alignment and clean facilities.
Reel-to-reel tape
- Capacity: also 5–10 MB per reel (depending on format).
- Access time: minutes to mount, wind, and seek.
- Cost: a fraction of disk, making it ideal for bulk data, backups, and batch workloads.
Bottom line: disks were scarce and costly, so most organizations used them only for the hottest, most valuable data. Everything else lived on tape.
The First Tiered Storage (and Early Automation)
This wasn’t just operator practice — it was beginning to show up in software. At CSIRO’s Data Automation Division in Australia, engineers pioneered Hierarchical Storage Management (HSM) concepts in their operating systems.
The logic was simple:
- Keep active data on disk for fast access.
- Migrate colder data down to tape automatically.
- Recall it back to disk only when needed.
That policy-driven placement was the ancestor of:
- Automated tiering in enterprise arrays (flash ↔ HDD).
- Software-defined storage balancing across media classes.
- Cloud storage economics — hot, infrequent access, and archive tiers.
And while today we extend those principles with machine-driven analytics and AI-based placement policies, the DNA traces directly back to those 1960s HSM experiments.
Networking & Accessibility
Networking still didn’t exist in storage terms. Your “SAN” was a cart, and your “protocol” was the operator rolling a tape or a disk pack across the room.
The step forward in the 1960s was media portability: instead of shutting down and reattaching entire cabinets, operators could swap disk packs. That was safer, faster, and less disruptive — but it was still logistics, not true sharing.
What changed was mindset: the idea that data could move independently of compute started to take hold. That seed would eventually sprout into NAS, SAN, and the Ethernet-based disaggregation we live on today.
What the 1960s Gave Us
- Random access became an expectation. Disks weren’t experimental anymore — they were business tools.
- Tiering was born. Economics forced a two-tier approach: fast-but-scarce disk versus slow-but-abundant tape, managed by both operators and early software.
- Operational friction dropped. Swappable packs reduced the downtime and risk of moving entire storage cabinets, setting a new baseline for usability.
The Bridge to Now (Why It Matters)
The 1960s gave us the economic and architectural logic that still guides storage today:
- Put the right data on the right media at the right time, driven by cost vs. performance.
- Reduce operational friction and you unlock new architectures.
Fast-forward 60 years, and the same logic drives flash tiers, object storage archives, and NVMe over TCP over fast Ethernet. We’ve traded carts for cables, and human mount queues for API calls — but the principle hasn’t changed.
Next Up
The Filing Cabinet Gets a Makeover: Winchester Drives, 1973 — how sealed assemblies bent the reliability curve and paved the way for SCSI, RAID, and the networked storage revolution of the 1980s.
This is the second post in a 12-part series charting that journey — decade by decade, technology by technology — showing how the cost-per-GB cliff, networking advances, and software innovation got us from washing machines to the frictionless, fabric-native storage we’re building today. To learn more, read the blog series:
