The latest I've read seems to indicate that for current generation SSDs, defrag (or any time of write) will have negligible impact on life. This stress test is up to 600 TB written to each of the drives they are testing: http://techreport.com/review/26058/the-ssd-endurance-experiment-data-retention-after-600tb
Cheers Ken From: [email protected] [mailto:[email protected]] On Behalf Of Micheal Espinola Jr Sent: Tuesday, 20 May 2014 7:14 AM To: ntsysadm Subject: Re: [NTSysADM] Defrag and SSD's I guess that depends. As you and Ben rightfully added: Backups. Anything else goes back to MTF and ROI, as defragmenting SSD's has been shown to significantly lessen the life of an SSD. I dont know if there are any statistics that show if a defragmented drive has a higher data recovery rate than a fragmented one for solid or mechanical drives.. Given the seemingly random nature of failures (although I dont know if there are commonalities for physical placement on disks), I dont know if this could ever really be quantified in a meaningful way. Maybe someone with more technical knowledge of flash-type memory could add something meaningful here. -- Espi On Mon, May 19, 2014 at 1:04 PM, Daniel Wolf <[email protected]<mailto:[email protected]>> wrote: Wouldn't a defrag make it easier to reconstruct data in a loss scenario where you're (geeze I can't think of the name, finding files by structure). Inb4backup Daniel Wolf From: [email protected]<mailto:[email protected]> [mailto:[email protected]<mailto:[email protected]>] On Behalf Of James Button Sent: Monday, May 19, 2014 2:40 PM To: [email protected]<mailto:[email protected]> Subject: RE: [NTSysADM] Defrag and SSD's Remember that a SSD is a cluster of storage cells that the management program in the SSD maps to the drive image that the OS driver sees. Defragging the drive from the PC OS does not physically sequence the data in those cells, it causes the storage management facility on the drive itself to go through the (to it) arduous process of copying cells of data to other cells, adjusting the physical to logical map, and then erasing the data from the cell 'cleared' Erasure and writing may possibly take 4 times as long as a read, and actually reduces the life of the 'erased' data storage cell as well as the storage used for the physical to logical map. Defrag originally had a triple benefit: Avoid head movement delays while disparate parts of the file were read - Does not apply to SSD's Avoid extra reads of the space allocation table required to assemble a list of the locations of the data on the drive - Not particularly relevant on drives with large cache, and systems with gigabytes of real memory Collect the FAT directory entries into a single block of storage of the drive (usually ordered, and clustered with other directory lists) - Certainly does not apply under NTFS where all the entries are assembled as detailed by MS and according to the space available within the MFT, and the file's creation full name. And I have been told that Windows will do some compression/tidy up work on the MFT as a background task So - I'd say defrag of a SSD is actually more likely to cause the system to be slower, than to be faster! But then again, that's only my understanding! JimB From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Micheal Espinola Jr Sent: Monday, May 19, 2014 8:00 PM To: ntsysadm Subject: Re: [NTSysADM] Defrag and SSD's Its never been a question if fragmentation happens on an SSD. It does. The question is whether or not its worth de-fragmenting: 1. Are you getting a worthwhile performance gain? 2. Is the effort required deteriorating ROI due to user interference or shortened MTF in a cost-prohibitive manner? Everything I have read indicates that it significantly hurts the MTF, and the performance gain is negligible. ROI of course can vary greatly depending on what you are doing with the equipment, and what that means to you fiscally. So, it is possible that defrag'ing an SSD is a worthwhile investment. I would imagine that is something akin to the way [fraudulent] investors short the stock market by paying for their analytical servers to be physically closer to the Exchange's servers. Thus gaining micro-seconds of advantage for every trade. -- Espi On Mon, May 19, 2014 at 11:42 AM, Dave Lum <[email protected]<mailto:[email protected]>> wrote: Truth, or salesmanship? http://www.condusiv.com/knowledge-center/videos/videos.aspx?index=13 I wonder if the actual productivity increase would more than pay for the product. Dave "skeptical"
