Samsung HD Unit Bought By Seagate

nanoflower followed up on a recent story about the unpredictable future of data storage. That story talked about Western Digital buying Hitachi, leaving just 4 players. Now: "Yet another hard drive company is going by the wayside, as Seagate is buying the Samsung HDD unit. Seagate is buying the unit for $1.375 billion (half in stock, half in cash)."

Well crap

[FrozenFOXX]

I've actually been a fan of my Samsung hard drives. So far they've outlasted every other drive manufacturer I've tried. Now I know that technically they all usually have roughly similar failure rates, but at least from personal experience right about every Samsung product of any kind I've bought I've always gotten great service on and great reliability from, something important for me with hard drives.

Seagate? Not so much. Well, guess it doesn't matter now as like it or not that's who we're getting. Still, I can't imagine a shrinking consumer drive market is very good for the consumer.

Consolidation of failures?

[Anonymous Coward]

Let's roll back the clock a bit
Maxtor (which made terrible drives) bought Quantum (which made good drives)
Seagate (which also makes terrible drives) bought Maxtor (which made better drives than pre-Quantum maxtor)
Seagate then starts making less terrible drives, and buys Samsung (which makes passable drives)
So the end result is we've had a consolidation of drive manufacturers which make low-end drives. Maybe that will squeeze some of the low-priced-low-reliability drives out of the market since they're no longer competing.

I'm not suggesting that segate and samsung intentionally make bad drives, but rather the drives they sell in the low-priced segment tend to be the loudest, slowest, least reliable drives I've ever had to deal with. One drive in my system right now, the error counter in the S.M.A.R.T system is incrementing by the thousands, where as the WD drives aren't incrementing at all. The drive seems to work, and isn't reporting that it's going to fail, but this just doesn't seem right.

Yes I see the inevitability of drives with moving parts disappearing, but not until a fundamental change in OS design happens.
1) No more swap files. This is the largest reason why we can't move to SSD's, because computers don't yet come with enough RAM, and OS's like Windows and Linux throw data into the swap file continuously. FreeBSD on the other hand you can have an uptime of 2 years and never consume any pagefile out of the box. They keyword here is "out of the box."
2) No more temporary files. How I see this working is that future "high end" SSD drives come with two partitions, a large writeable partition that is directly writeable, and a smaller RAM based copy-on-write partition that only commits changes to the NAND upon shutdown or power loss. In *nix'isms this would be the /var/run , /tmp and maybe the swap partition. Low-end drives would omit the BBU and have smaller/slower RAM, so that accidental power loss would just wipe the pagefile/temporary storage, and tout it as a security feature.
3) Changes in filesystem design to support wear leveling. None of the current file systems are any good at this, particularly with journaling. The best I could see happening is that all the OS manufacturers agree to support a single file system standard for NAND devices, unfortunately that's probably not what's going to happen. The problem with current file system's is the need to change so many bits uselessly (eg Access time and Modify time) while doing absolutely nothing to the file. Every time you "Search" for a file you end up wearing down every file on the hard drive. This has got to stop. As with #2, file modification/access time's need to be stored in a RAM section of the drive and written only when shutting down.