2

I tend to download about 100GB stuff every month and most of it over torrent and I have been doing this for almost 2 years.
I download things to my windows drive and then move them to other locations.
I was wondering how would this affect my hard drive's health because I have read that HDDs have limited read write cycles.
Furthermore, I have observed that torrent clients write data quite randomly onto disk(by randomly I mean scattered/fragmented).
Will this affect my hard drive's health?

2
  • Where did you read that HDDs have limited read/write cycles?! Jun 28, 2012 at 5:53
  • Don't remember....it was a long time back...seems to be wrong after a few google searches.. Jun 28, 2012 at 6:23

2 Answers 2

10

Actually, HDDs have mean times between failure - they tend to usually die due to mechanical failure or electrical failure, and this will happen anyway, due to run time, not data transfer.

I suggest taking a look at this old closed question of interest on hard drive lifespans here - which seems to reference this paper by google - they have a graph of annualised failure rate for 5 years on page 5. It indicates, outside the first year (when the obviously defective disks die) and the 5th year (by which you may want to consider replacing the drive), utilisation has very little effect.

SSDs have limited read/write cycles, and probably arn't a good choice for torrent storage

Fragmentation can be an issue, but newer versions of windows defrag automatically, and its a performance issue rather than a lifespan issue. I believe many torrent clients have an option to preallocate the space to avoid that as well.

I personally ended up having a seperate drive for OS, transient storage and long term storage, with the transient storage drive including the long term storage drive's backups. On a longer term, i may look at building a san for this.

A safe idea is to assume any drive can fail at any time, and work accordingly, but I don't really think that downloading torrents has a greater effect than the fact that your hard drive is running or the temperatures its running under,

2
  • The two Hitachi HDS728080PLA380 80 GiB HDDs with software Raid1 / Raid0 that I used for daily work (OS + storage) and 24h 7/7d torrenting since 2005, are still doing fine as my father's Ubuntu desktop's main drives. For three years, that add to the 1869 days I had them with max temp 52°C (126 °F). Had rtorrent configured for whole file pre-allocation and delayed write. Hard drives are here to serve. Torrenting and desktop I/O are little compared to server / database's.
    – tuk0z
    Oct 5, 2015 at 10:41
  • Indeed. Most hard drive failures are on the electronics board and are not related to "numbers of IOs" but rather to total power-on time. Mar 16, 2019 at 0:46
4
  1. Applications do not determine where they write on the disk. THat is up to Windows which uses the Master File Table ($MFT) a super hidden file on the disk that keep track of allocated blocks. So when an application such as explorer or winrar or an installer requests disk space, Windows takes care of which spaces are free from the beginning of the volume and fills them in that order. $MFT for NTFS replaced the previously used similar function with $FAT32 File ALLOCATION sp. Table.

  2. Constant thrashing of files write/install/delete/move leaves your drive extremely fragmented since each file is unlikely the same size as the ones deleted. Find a good defragger that allows maybe allows you to specify zones for big files. Otherwise just use Auslogics free defragger or SysInternals free defrag tool as a scheduled task. It is not critical to do that often as Windows is multitasking and doing random seeks anyway with registries, cache and %temp% files , but monthly may be ok if you want to see your files more contiguous. Just add a task to Schedule task point to the defragger and set a schedule.

  3. Otherwise disk drives ought to last 5 yrs non-stop trouble-free. YOur mileage may very depending on if it is dropped in an external box when running or a laptop or is running hot in a tower with other drives stacked too close and no air-flow between them.

4
  • quite nice answer....especially the first point....and thanks for the other tips...but i am inclined towards giving credit to Journeyman because his answer directly addresses the query that whether data writing will affect HDD health... Jun 28, 2012 at 6:20
  • No Problem. I spent 11 yrs in the HDD engineering business as Test Eng. Mfg for OEM disks in the 80/90's and Unisys Peripheral manufacturing. But since failure modes for every vendor may vary depending on weakest link, I didn't want to bore you with all the root causes of failure, but writing is a weak link for SSD not HDD. Maybe you read that and were thinking HDD. Anyways flash write algorithms randomize blocks so usage is distributed evenly and that extends MTBF of SSD from 10K write cycles 10~1000x depends on % free space. The weakest link for HDD's are shock, heat and moisture Jun 28, 2012 at 11:36
  • The thing about MTBF now is that even SMART technology is useless since ECC has become so powerful and in use since the age of high density. I used to reverse engineer everyone's drive to qualify them for OEM corp purchases and in those days as now "user experience and average MTBF" did not match due to variance in handling and environmnent. Now there is no almost warning for failures whereas before soft error rates were an early warning. ECC has raised the bar so they rarely exist. Hitachi had the best MTBF in high end$$ drives in Japan when Unisys was the #1 computer company in Japan. Jun 28, 2012 at 11:52
  • back in the 80's write error rates were on the order of 10^12 and depended on vendor's thermal compensation for overshoot due to magnetic fatigue with heat after servo calibration on power on. Overshoot on random seeks occasionally caused it to be written off track if the sector just happened to be next when a write after seek completed. I discovered this on Toshiba drives and Japan's Eng Mgr came for a month to Winnipeg to fix the design at our Burroughs Facility. ( no longer in existence ) But that was with a dedicated servo surface and stacked heads& disks. Now its embedded servo per sector Jun 28, 2012 at 12:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .