SSD Longer lifespan than traditional HDD? True?

Determining life expectancy of a chipset is a bit different from a harddrive. The life expectancy of an SSD has multiple dependencies. According to the guys on XS, file size can impact life expectancy. They also seem to be judging the life expectancy of SSDs in general based on a few SSD models. The results from their testing, being done on so few drivers, makes the results inconclusive. If they were using a batch of say, 1000+ drives, their "testing" would hold a bit more weight. They're also using systems that are powered on 24/7...whereas simply applying and removing power can impact the life-span of electronics devices due to component quality. Makes little difference how long the NAND can survive read/write operations....if the controller can't handle the system being shutdown and booted regularly... I know....most tests don't include cold booting the system....but, I've seen more harddrives fail at boot than during use.

 

Thats not quite true, the expected life expectancy is longer than the warranty period, they don't want a whole heap of returns a few months before the warranty runs out!

I'd say the statement is true for a worst expect case scenario, where there is high temperatures (still safe though), medium use, realistic star/stop cycles etc.

If you run the drive 24x7, having the power features turned off can actually be beneficial. Not for power use but for longevity. The most strain on the drive occurs during spinup, where the drive has to get up to 7200/5900/5400 rpm within a second or so, so you can access your data.

A hard drive when spinning at speed only needs enough effort to overcome the extreme little bit of friction there is to keep spinning, and in these cases as long as its reasonable quality it could conceivably spin a very long time. Few platters are better. Remember that the platters in the drive are inside a case, and that the platters are extremely well balanced. This is unlike a case fan, as the case fan has a lot more effort required as it has to not only spin the blades but push the air as well, and is most likely not as balanced as hard drive platters.

The life expectancy of an SSD is related to temperature, write cycles etc, as you are well aware. The thing with SSD's though is people sometimes try and install it in low memory systems, such as 2 GB, which you can imagine the resulting pagefile then wouldn't play nice with the drive. Ideally for an SSD you have 8gb of RAM. Windows is designed to operate differently with SSD's present, having at least 8GB of RAM will ensure minimal drive write cycles usage and best performance. 4GB is not quite enough to be on the safe side.

If you just had the SSD sitting there and not doing much reading and writing, then the life expectancy can conceivably very easily be many decades!

For a mechanical drive, and this goes for anything else, if they say 100,000 start/stop cycles, and 20 years expectancy, they don't mean you can get 99,995 start/stop cycles and 19 years expectancy and still have the drive running. Each aspect of the life expectancy of an item is inversely proportional to each other aspect of the life expectancy. For a mechanical drive for instance, the more start/stop cycles that are done, the fewer years you can expect the drive to last. Conversely, the longer you have the drive running, the few start/stop cycles you will be able to do before the drive is expected to fail. This is why the ratings sound so good but in reality they aren't. You aren't going to have one start/stop cycle after 10 years, nor are you likely to have 10 'on years' if the drive turns off with power management . The greatest risk to a drive in its operational use though is its start/stop cycle, so that should be considered.

So, for longevity of mechanical hard drives:
- keep vibration etc to a minimum
- keep the drive cool
- turn off the power saving for HDD's and Windows so they don't keep turning off and turning on
- ensure the power supply to the drive is 'clean'. This is pretty easy with a good quality power supply and SATA cables.

Longevity for SSD's:
- keep cool
- Minimise write cycles by having enough RAM (at least 8gb) so that pagefile use is kept to a minimum
- don't let the drive get overly fragmented. Despite what they try and claim performance etc does suffer. You need to use an SSD optimised defrag program (such as Perfectdisk 12), NEVER use a normal defrag as they are optimised for mechanical drives.
- Clean power supply. Again this is pretty easy with a good quality power supply
- To the extreme, you should also minimise vibration. Sure, SSD's are pretty safe against vibration, but there are different frequencies and amplitudes of vibration and you never know if you are creating resonance inside any component of the drive

Resonance can be very bad
http://www.youtube.com/watch?v=3mclp9QmCGs

 

I think you meant Diskeeper 2011 with HyperFast. Although PerfectDisk 12 can optimize an SSD too, Diskeeper adds the very unique benefit of running its tasks in the background without slowing the system down at all (not just on SSDs but on harddisks also) - whereas PerfectDisk is actually known to hog system resources alot more than really necessary.

If you have an SSD then HyperFast is the way to go IMHO. You can just leave all of Diskeeper's features enabled 24/7 without ever causing a negative impact on system performance, it's brilliant innovation.

 

Hopefully Diskeeper 2011 is a bit better than previous versions, the last couple of times I tried diskeeper it ended up making the drive slower! (from being defragmented with perfectdisk), and yes I did uninstall Perfectdisk first. It could have been something specific to those versions though.

That said, Perfectdisk 12 is better than previous versions and I don't think the 'resource hog' situation applies anymore. Perfectdisk also has a system now called 'optiwrite' and a few other things, but it too is better for SSD's.

Anyways, the SSD's requiring no defragmentation is a myth probably created by the earlier SSD marketing teams, or by some reviewers who didn't know what they were talking about. Fragmentation can create a several fold slowdown in performance. Its just if the drive is normally at 500 MB/s, people don't notice it when it is only going at a slow, miserable 200 MB/s! Actually fragmentation can reduce it significantly more than that.

The same people who say SSD's don't need defragmenting are probably the same people who still have the bios set to IDE-compatibility mode! A no-no for mechanical drives and a definitely no-no for SSD's! and a little annoyance to change to AHCI after Windows is installed. Basically Microsoft disable unused drivers, one of which is the msahci driver which is required to be loaded for AHCI mode (otherwise you will BSOD on boot). If enabled and in IDE mode in the bios, it takes up no RAM, but the slowdown of loading a massive 31kb file (on Win 7 x64 SP1) is just too much, right?!

It was pretty much an oversight by Microsoft years ago, they put it in a kb article years ago, and they still haven't made it default even though it should be in Vista and Windows 7.

So, common oversights/mistakes:
- Not enough system RAM
- Fragmentation
- IDE mode set in the bios, which is a pretty much default scenario
- Use of not the latest Intel RST as most versions don't support TRIM. Not sure about AMD. Latest Intel RST (includes the AHCI driver) is currently 10.6.0.1002

 

Technically why would you need to defragment your SSD?
My boot SSD is 32percent fragmented but I have not experienced any slowdowns

http://www.diskeeper.com/hyperfast/

No mention of newer SSDs running on newer OS (Windows 7) Sounds like it may be related to a lack of TRIM in these older systems.

 

You don't actually want to "defragment" it the exact same way you would with a normal harddisk. Instead, you want to use SSD optimization software (that usually comes with defragmenting software as a separate feature, sometimes already included and somtimes optional).

The SSD optimization software consolidates free space (to eliminate write amplification, which is responsible for SSD performance degradation) rather than relocating clusters to convert fragmented files into contiguous files.

 

I don't know which older version you are referring to but if it's an early build of 2009 then yes, maybe I can understand. Automatic defrag mode and IntelliWrite have been working flawlessly together for me since version 2010 on both my laptop and my Intel Atom based netbook (I don't use manual defrag because I leave automatic defrag mode enabled at all times - even while gaming, "fire and forget"-style).

 

Actually, most military equipment until about a year ago still used HDD's. I remember using the Panasonic Toughbook, that had a HDD. But I'll be damned if we could break that damned laptop; we dropped it, spilled liquids on it, left it in the boiling sun, smashed the screen a little hard and do nearly every single thing to it but shoot it, but the damn thing kept on ticking.

 

Because maybe you are confident to the slow speed. The slowing has maybe come over time so you didnt notice ?

 

Reads dont slow over time due to fragmentation.
There is no seek head so you dont have to wait for it to move.
Therefore it doesnt matter where the data is.

 

I like the idea, but I've never had a HDD fail, SSD's not been out long enough to say (synthetic read/write tests are meaningless).

Also I need 256GB or above and won't pay more than £80 so will wait a while, but if you look at my specs I already have 9.7TB and will be adding another 2TB in the coming weeks (although lots of duplicates but cba sorting it). But even without my downloads with all my apps and games I'd need 256GB.

 

My former employer used a Panasonic ToughBook....one of the guys at work actually drove over top of it in a Ford F350(it was closed at the time), then picked it up and finished what he was doing. The only damage ever done to it, was from guys jerking the USB cable out of the side....which finally broke the USB ports.

 

http://downloads.diskeeper.com/pdf/Optimizing-Solid-State-Storage-with-HyperFast-Technology.pdf

 

The latest version of HyperFast (with Diskeeper 2011) actually focuses on Windows 7 also.

https://www.diskeeper.com/partners/channel/products/hyperfast.aspx

TRIM is not supported on an SSD if the SSD is a member of a RAID array (although if Intel ICH9R / ICH10R is set to RAID mode in BIOS, TRIM is supported on an SSD connected to it if the SSD is not a member of a RAID array).

It's worth notingt that, even though a special Windows command sequence can be used to confirm that TRIM is properly enabled in Windows, there is no guarantee whatsoever that TRIM is supported by the SATA controller itself. In fact Marvell SE9123 / SE9128 did not support TRIM until a new driver was released recently.
Marvell SE9123 / SE9128 did not support TRIM: http://forums.crucial.com/t5/Solid-State-Drives-SSD/Marvell-9123-9128-TRIM-yes-no/td-p/11109
Marvell SE9123 / SE928 now (finally) supports TRIM: http://forum.crucial.com/t5/Solid-S...Marvell-91xx-drivers-supports-TRIM/td-p/44440

EDIT: Scratch my remark about Marvell SE9123 / SE9128 supporting TRIM, reports are still totally inconclusive.

 

Good info cheers, a point well made.

How do you rate the performance increase in %? (if you can put a figure on it)

 

It varies greatly IMO depending on the make and model of the SSD, usage scenario, and TRIM yes / no.

I think it can make quite a bit of difference especially on older SSDs with not much free space on them if alot of file deletions / creations are going on but I can imagine even newer SSDs based around SandForce can also benefit alot from SSD optimization software if TRIM is somehow disabled so if TRIM does turn out to be unsupported by Marvell SE9123 / SE9128 after all (see my previous post) then I guess at least some OCZ Vertex 3 / Agility 3 owners might still feel like giving it a shot because of how Garbage Collection works on SandForce. I suspect Crucial C300 owners would not see a difference because the C300 uses a dual core ARM processor with its own OS to handle Garbage Collection, which causes TRIM versus no TRIM to hardly matter at all. I'm not sure about any of all this but IMO it would be interesting to find out.

 

Thanks for the insight.

 

That Diskeeper nonesense is all bull****. uke2:

It applied to old SSD´s which did not have any form of build in garbage collection algorithm.

All modern SSD´s since the Intel G1 or Indilinx Barefoot have that function in their firmware.

 

In answer to the thread title - no, at least in my case.

I'll explain.

I purchased a 128 GB Crucial M225 SSD back in late December 2009 to use as a boot drive solely for Windows 7 64-bit and non-game applications such as Paint.NET, etc, etc. I installed it between the Christmas and New Year period, switched my BIOS to AHCI and it ran flawlessly. I did some benchmarks, updated the firmware to a newer one that supported TRIM and enjoyed my purchase. Windows felt uber-responsive, loading times were quick and I couldn't be happier.

Like many I'd imagine, once they find out SSDs have a limited number of writes, I reallocated my music, documents and videos folders to a separate SATAII hard drive (I've always done this anyway in case I should lose the C: partition through corruption plus it made the weekly Acronis True Image backups smaller). Basically, I limited how much data was written to the drive in order to preserve it's life. The SSD had no page file (those were on the other three hard drives), defragging was disabled as was superfetch, etc., etc. Yet 18 months later my SSD is at 18% of its remaining life with SSDlite Free warning me that it has less than a month of usage left. I believe it has dropped 5% in the last month or two from memory so it isn't going to see another year at the current rate.

The weird thing is that prior to updating to firmware v2030, the remaining life was at around 80% but it dropped to 40% after I'd installed it. I've no idea why. Possibly the firmware has been rewritten to be more conservative with its estimates but it is still disturbing all the same to see an estimate of how much life my £300 drive has.

I've contacted Crucial about this as the drive has a limited five year warranty (I wonder what the limited part refers to?) and they claim a life span far in excess of what mine appears to have. CrystalDiskInfo shows that the drive has 5,500 hours of use.

Goodness knows what would have happened had I carried on benchmarking the drive (those caused a 5% drop from 100% to 95% before I stopped) or if I'd installed/uninstalled lots of games on it.

I am planning on buying a totally new PC toward the end of the year with SATAIII and USB 3.0 once Intel release the new high end processors with quad-channel memory. I'll likely buy a new one then but there's no way I could ever go back to not using an SSD so my new one will definitely have one. I only hope that these have longer lifespans because I typically buy a brand new PC every three years so realistically I'd want an SSD that would last that long.

Can they last three years? Based on the model I have, I'd say they don't so that makes an already expensive drive even more so if you have to replace it with another after 18-24 months.

 

And now you are worried about some obscure numbers. I hope you did not wet you pants. LOL.

Clearly the amount of wear did not change or accelerate with that firmware flash. No need to worry about anything.