Fullscreen HTML5Med Res (??MB)HTML4

SILICON SILIC CHIP www.siliconchip.com.au Publisher/Editor Nicholas Vinen Technical Editor John Clarke – B.E.(Elec.) Technical Staff Jim Rowe – B.A., B.Sc. Bao Smith – B.Sc. Tim Blythman – B.E., B.Sc. Advertising Enquiries (02) 9939 3295 adverts<at>siliconchip.com.au Regular Contributors Allan Linton-Smith Dave Thompson David Maddison – B.App.Sc. (Hons 1), PhD, Grad.Dip.Entr.Innov. Geoff Graham Associate Professor Graham Parslow Dr Hugo Holden – B.H.B, MB.ChB., FRANZCO Ian Batty – M.Ed. Phil Prosser – B.Sc., B.E.(Elec.) Cartoonist Louis Decrevel loueee.com Founding Editor (retired) Leo Simpson – B.Bus., FAICD Silicon Chip is published 12 times a year by Silicon Chip Publications Pty Ltd. ACN 626 922 870. ABN 20 880 526 923. All material is copyright ©. No part of this publication may be reproduced without the written consent of the publisher. Subscription rates (Australia only) 6 issues (6 months): $70 12 issues (1 year): $127.50 24 issues (2 years): $240 Online subscription (Worldwide) 6 issues (6 months): $52.50 12 issues (1 year): $100 24 issues (2 years): $190 For overseas rates, see our website or email silicon<at>siliconchip.com.au * recommended & maximum price only Editorial office: Unit 1 (up ramp), 234 Harbord Rd, Brookvale, NSW 2100. Postal address: PO Box 194, Matraville, NSW 2036. Phone: (02) 9939 3295. ISSN: 1030-2662 Printing and Distribution: Editorial Viewpoint Solid-state drive pitfalls As we increasingly favour solid-state drives (SSDs) over traditional hard drives for their speed and reliability, it’s crucial to understand their limitations. While SSDs have transformed data storage with their efficiency, they are not without pitfalls, particularly regarding long-term data retention. Common wisdom suggests that SSDs are not ideal for archival purposes. Data written to an SSD that’s left unpowered for extended periods is at risk of corruption. Regular usage is essential to avoid data degradation. A less well-known problem occurs even if your SSD is powered up daily, affecting many different brands and models of SSD. It may not affect all of them; some could have mitigation strategies. However, I have experienced it with a couple of different brands. This problem occurs when you write data to an SSD and then don’t access it for a long time (months or years). It happens even if the drive is actively used, as long as that particular data is not touched. When you go to access it again later, it is very slow to read back. While freshly written data may read back at 1000-2000MB/s, after a few months or years, it might only do 50MB/s. Some reports I found from other users said that their drives barely managed 5MB/s! That’s not only a lot slower than the SSD with freshly written data, it’s much slower than even an ancient mechanical drive. After searching the internet, I only found a few reports of this phenomenon, far less than I expected. I attribute this slowdown to voltage drift in the flash cells. For example, SLC flash stores one bit per cell as a voltage level. Over time, that voltage can shift closer to the point that distinguishes a ‘zero’ from a ‘one’, narrowing the margin for error and necessitating slower read speeds to ensure accuracy. Like DRAM, I suspect that the ‘sense amplifiers’ used to convert the analog voltage levels into digital data in a flash chip have a ‘settling time’, and that time will be more extended as the margin between the cell voltage and the threshold narrows. Therefore, the controller will automatically throttle reading back to a slower speed if it detects too many errors. The challenge is more pronounced in more common multi-level cell technologies like MLC, TLC, and QLC. With their finer voltage distinctions, these flash devices are even more susceptible to drift, necessitating extensive error correction if the voltage drifts and, consequently, slower read speeds. I also suspect that when the SSD controller reads back cells with voltages that have drifted significantly, it will be programmed to write that data back to refresh the cells, avoiding data corruption. That will also slow down reading. SSD controllers could be programmed to periodically refresh data in the background, mitigating voltage drift. However, this feature seems lacking in many models, as evidenced by widespread slowdowns. For now, manually refreshing the drives may be a necessary workaround. Software is available to do this automatically, reading back all the data on the drive and rewriting it. It would need to be run periodically, eg, every few months, to avoid slowdowns. I’d like to hear if others have encountered similar problems with their SSDs. Have you noticed a significant slowdown in SSD read speeds over time? Are you aware of any other strategies to counteract this problem? By the way, we’ve added some information on this phenomenon to our article on Computer Storage in this issue as it seemed appropriate, given that it specifically discusses flash memory technology. Still, I thought I would expand on it here, giving the issue more attention. Note on Vintage Radio: you may have noticed that recently, we have been indexing Vintage Radio columns on the Contents page along with other articles instead of in the sidebar. However, nothing has changed in the column itself; it is still ongoing. A question for readers: given that the column often describes nonradio vintage equipment (amplifiers, test equipment etc), should we change its name to just “Vintage”, or perhaps “Vintage Gear”? by Nicholas Vinen 24-26 Lilian Fowler Pl, Marrickville 2204 2 Silicon Chip Australia's electronics magazine siliconchip.com.au