SSDs Suffer Performance Degradation Over Time

Rob Williams

Editor-in-Chief
Staff member
Moderator
From our front-page news:
We've talked a lot about SSDs in our news section before, and for good reason. They're small, fast and desirable in almost every way. But, like most things that are "too good to be true", there are a few caveats - namely price and fragmentation. Our friends at PC Perspective decided to thoroughly investigate the latter issue, and I have to admit, the results are very interesting. Simply put, while SSDs can offer incredible performance at first, their performance can degrade fast.

Due to how SSDs are designed, they're bound to become fragmented, and that issue is amplified when we're talking about models with small densities, such as 80GB or lower. Fragmentation might sound like a simple issue, but really, the effect it has on performance is huge, and the problem goes beyond something that a defragmenter could fix (most often they will make things even worse). In some of the site's examples, the write speed degraded to around ~25MB/s, from the normal 70 - 80MB/s. That's massive. Surprisingly enough though, that issue can actually be remedied...

... but, with yet another downside. In order to restore the drive to its original performance state, you'd have no choice but to format the entire thing. If the drive is stores your main OS, you can likely trick it in a few ways to improve performance, but you'll never see the performance you did after the first OS install. The article goes into depth as to why, and how to remedy it, but in truth, it really doesn't put SSDs in a terrific light. Larger SSDs would put the issue off for a little while, but that would all depend on how fast you fill it up. Hopefully OS' and SSD memory controllers (or the firmware) can become even smarter sooner than later to avoid such performance degradation in future models.

intel_ssd_performance_degredation_pcper_021609.jpg

Once internal fragmentation reached an arbitrary threshold (somewhere around 40 MB/sec average write speed), the drive would seem to just give up on 'adapting' its way back to solid performance. In absence of the mechanism that normally tries to get the drive back to 100%, large writes do little to help, and small writes only compound the issue by causing further fragmentation.


Source: PC Perspective
 

Kougar

Techgage Staff
Staff member
I'm glad you posted this. Amazing how none of the major sites seem to have touched on this issue yet.. One or two sites mentioned this occured but not why nor had any mention of the severity.

Frankly it's very disheartening because there is a secondary issue with Intel SSD's that makes this an even bigger problem for Intel SSDs, which are generally regarded to be the best SSDs out there.

Buying an X-25-M then having it perform worse than a 5400RPM laptop drive makes it just seem like a huge waste of money. I hope they update the ATA specs quickly.
 

Rob Williams

Editor-in-Chief
Staff member
Moderator
Kougar said:
Amazing how none of the major sites seem to have touched on this issue yet.

I'm wondering if it's because not many editors have actually used an SSD as their main system drive? For example, I have two SSDs here, but have never experienced the issue, but it's because I don't use it as a main system drive. I format the drive, restore a Ghost image and go from there. The only output I really have after that point, other than Vista's doing, are log files from FRAPS. Even with 20GB of space free, I've never noticed a speed decline, but I admit I've never really thought to run HD Tune on it either, at that point.

That aside, I agree... to see the performance drop like that is without question, ridiculous. For a storage solution that's already much more expensive than a typical hard drive, it should always be faster. The fact is, the only way to restore the performance is to format the drive later, and if it's your OS drive, which it probably would be, that's a huge problem.

SSDs are still relatively in their infancy, but we need to see better specs fast.
 

Kougar

Techgage Staff
Staff member
I'm guessing that might be the case. If they are always writing a single large installation image to the drive before every series of benchmarking, then fragmentation of the LBA is going to take significantly longer to occur.

I'm more curious when they will get around to updating the ATA spec with the new commands that will allow the drives to selectively delete LBA data. That could be measured in years, or come with the SATA 6Gbps spec, or who knows when...
 

On_Wisconsin

Coastermaker
I've had my SSD working since November, no significant degradation with about 40GB of 60 used. Albeit this is not the gaming computer, so only startup time and Explorer response time is only what matters. Still boots, reads faster than my main computer with a 7200RPM HD...
 

Rob Williams

Editor-in-Chief
Staff member
Moderator
Kougar said:
I'm more curious when they will get around to updating the ATA spec with the new commands that will allow the drives to selectively delete LBA data.

I agree, but the sad reality is that these things can take FOREVER, which truly sucks. The technology as it stands is so good, and it's "almost" perfect, but this spec is sure to keep us back a little bit.

On_Wisconsin said:
I've had my SSD working since November, no significant degradation with about 40GB of 60 used.

Yeah, usage patterns come into play here, and chances are many people wouldn't notice a speed decline either. That's far from being a bad thing though. I think the worst of it would be with constant writing of log files or even small files from IM chats and such. It might take a lot longer for some people to realize that something's wrong.
 
Top