CKS blogged something to today that caught my eye, in the context of filesystems defined in host byte order:
In the beginning, storage was close to 100% system specific. Not only did you not think of moving a disk from a Vax to a Sun, you probably couldn't; the entire peripheral interconnect system was almost always different, from the disk to host cabling to the kind of backplane that the controller boards plugged into.
Although the above could be the case in the West, in USSR it was quite common to move removable media between systems, even back in the 1970s. The cartridge of choice was the disk pack for IBM 2311, a 7.25MB, 11" hard disk stack of 4 or 5 platters. It could be easily transported between BESM-6, IBM/370 clones (ES-1022 and up), PDP-11 clones (SM-3, SM-4, SM-1420, etc.), HP 2000 clones (SM-2), and Mitra-15 (ES-1010). It only went out of service by about 1986, with the adoption of 29MB disk packs.
Granted, UNIX only worked on PDP-11. It reached ES series too late for the 7.25MB packs, in the ES-1045 generation. However, SM-4 brought an "RP-5" cartridge with a single platter, which for a while was a gold standard for minicomputers. In Russian practice, it used a half-density recording for 2.5MB instead of 5MB in western cartridges. The "RP" ("rk" in UNIX) drive was hooked to a wide variety of mini- and micro-computers during their brief popularity before they were supplanted by PCs. Aside from the original SM and smaller LSI-11 compatibles (DVK), it was connected to Iskra-226 (Wang micro), Videoton's Z80-based TRS-80, Mitra-225, and basically any and every microcomputer, mostly based on the 8080 clone KR580IK80. The most popular format, I suspect, was a trivial filesystem used by DEC's RT-11 OS family.
When PCs came about, it was rather common to move around their hard drives with MFM interface, although Russian domestic winchesters required extreme care. Unplugging one with unparked heads could easily scratch the surface. Soon, the imports flooded the market and the 20MB Seagate ST-225 became the gold standard. It lasted until the single-cable IDE replaced it. Interestingly enough, the last hold-overs of LSI-11 line used various trivial IDE controllers with CPU-controlled access. You could attach something like ST-157 to them.
Chris was building an argument that the lack of portable disk media was the reason why everyone made their Berkeley filesystem in host byte order. He's not necessarily incorrect about that. Even if Russians used cartridge and winchester drives pervasively, nobody cared about them and they were not setting the software standards.
BTW, since we're on topic, in case of Linux, the order independence was not easily won. The original port to Amiga (by Geert, IIRC), used a big-endian ext or ext2, if not minix even. When SPARC port came about, DaveM was not sure if to follow Geert at first (again, IIRC). Someone made an argument that byte-swapping would waste CPU, and it carried some weight. Also, adding macros into all the correct positions were challenging. I remember arguing for order independence, in part because I did have an easy access to PCs with SCSI HBAs. Not that anyone listened to my opinion, but eventually DaveM and Tytso went with it as well, and the rest is history.
