• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

If your job depends on it, mechanical or SSD?

N

nk215

Senior member
Hello everyone,

If you can’t afford down time, would you use SSD as an OS drive on your computer?

That’s my dilemma, at home, most of my computers use SSD as boot drive to gain some speed. But at work, after thinking hard and long, I picked mechanical HD for OS and SSD for data (which is religiously backed up to 4 places, two on site and 2 off sites. The data is also backed up onto the mechanical OS drive). As long as the computer can boot into windows, I have access to the data I need.

RAID 1 or 5 (for boot drive) is not an option at the moment

Mechanical HDs are obviously slow compared to SSDs but their reliable track record for is known (except those new 2TB+ drives). SSD is still rather new.

Performance is not that critical since the computer s are not doing anything special. Office, IE, excel etc. The computers are Dell optiflex 7010 with the latest i7 cpu.

Thanks
 
If you choose a reliable SSD (such as Samsung SSD 830), there is no need to worry about reliability. Of course there is still a chance that the drive will fail prematurely but that is the case with all electronics.
 
If you can't have any down time and performance doesn't matter at all and isn't doing anything intensive, then I'd just go mechanical with RAID 1 and make sure they're setup correctly for hot swappable.
 
Even if the worst SSDs have failure rates similar to HDDs. The best SSDs are more than an order of magnitude better in terms of reliability. Both are unreliable so how you treat them in terms of backup and failure is identical so the minor difference in reliability between the two wont change your overall productivity. The speed difference between the two technologies however will.
 
SSD & backup, if its a desktop/laptop run windows backup or some other backup, assuming you have w7, it'll backup to your schedule... just get a big ass file server
 
rsutoratosu said:
SSD & backup, if its a desktop/laptop run windows backup or some other backup, assuming you have w7, it'll backup to your schedule... just get a big ass file server
Click to expand...

I agree. This is what I do. Both for web server and home desktop. Keeps things simple. And yet reliable.
 
Guys,

Raid is not an option at the moment.

Didn't anyone notice how often users report that the SSDs drop off in BIOS/BSOD/Boot Drive not found within the first year?

On paper, SSD is more reliable but it hasn't proven out in practice yet. I've experienced 2 mechanical HD failure in the last 15 years or so. Both of which were 5+ year old drives. I have no issue switching out the drives after 3+ years (computers get refreshed about that time too).
 
If you have that kind of backup system setup, you can use whatever you want for the OS drive.

Didn't anyone notice how often users report that the SSDs drop off in BIOS/BSOD/Boot Drive not found within the first year?
Click to expand...
Don't buy an OCZ drive. For business, Samsung or Intel only.
 
I would never touch an OCZ even if it were the same price as a mechanical HD. The one that I think of has some BSOD and BIOS drop off reports is the Crucial M4 (which I have 3 of w/o issue so far), from newegg.

I'll look into raid 1.

The only study that I found so far when they have thousands of SSD in a data server (155,000 intel SSD). In that case, the SSD is far exceeding the failure rate of the mechanical HD. However, these are Intel X25 SSD (both the consumer and the enterprise models).

After thinking harder and longer, I decided that I'll go with SSD but with the enterprise class devices. That means Intel X25-E 64G for the OS and Intel 710 for data if needed. I only need about 15GB for data so I may even be able to use an X25-E for that.

It's interesting to see that an enterprise SSD is still $10/G after 3 years.
 
Last edited:
nk215 said:
I've experienced 2 mechanical HD failure in the last 15 years or so. Both of which were 5+ year old drives. I have no issue switching out the drives after 3+ years (computers get refreshed about that time too).
Click to expand...

The process of manufacturing mechanical drives nowadays (in my opinion) has gone downhill since the days of the drives you have had. I agree with GlacierFreeze regarding the setup you will need.
 
nk215 said:
I would never touch an OCZ even if it were the same price as a mechanical HD. The one that I think of has some BSOD and BIOS drop off reports is the Crucial M4 (which I have 3 of w/o issue so far), from newegg.

I'll look into raid 1.

The only study that I found so far when they have thousands of SSD in a data server (155,000 intel SSD). In that case, the SSD is far exceeding the failure rate of the mechanical HD. However, these are Intel X25 SSD (both the consumer and the enterprise models).

After thinking harder and longer, I decided that I'll go with SSD but with the enterprise class devices. That means Intel X25-E 64G for the OS and Intel 710 for data if needed. I only need about 15GB for data so I may even be able to use an X25-E for that.

It's interesting to see that an enterprise SSD is still $10/G after 3 years.
Click to expand...

See, here, you're talking about how your job is on the line and how important reliability and availability is, but you're designing TWO single points of failure into the solution. That's asking for it.

1) Do not put your data on your boot drive.

2) RAID-1 mirror your data drives for availability.

2a) RAID-1 mirror your boot drives for availability.

3) If you can wait a month or two, go with the Intel SSD DC 3700. Intro price on the 100GB drives is 40% less than the current newegg price of the SSD 710.

4) You don't really need an SSD boot volume on a server that never reboots. You can probably "get away" with using mirrored spinners for the boot drive and mirrored SSDs for the data. Assuming end user I/O performance actually matters, and you have gigabit ethernet and all that jazz.

(Although price/GB calculations get thrown out the window with a data set that small. The least expensive SAS drives and the least expensive enterprise-grade SSDs are around the same price. The HDDs will hold 10x the data, but either will be way more capacious than you need for 15GBs of data.
 
Last edited:
Dave;

I never said I put the data onto the boot drive. If/When I go with the X25-E then it would be two X25-E drives. I didn't state that clearly in the above post.

I'll take your advice and wait for the new DC 3700. Thank you.

I think about RAID 1 a lot lately, it comes down to this:

1) I don't trust the built-in raid - had bad experience with it before
2) I can get a hardware raid card. That's not the problem but I face similar question: what's the chance of the card itself fail vs. the X25-E SSD fail?

Up-time is only as good as the least reliable part. If the raid card is not as reliable as the X25-E then what's the point of using it at all?
 
nk215 said:
Up-time is only as good as the least reliable part. If the raid card is not as reliable as the X25-E then what's the point of using it at all?
Click to expand...

Any computer part has a chance of failure. If you keep second guessing then you'll never build a computer. Have to just take the plunge and go for it. Buy spare parts if it really is that huge of a deal. Have an extra hard drive/SSD ready to go so you can hot swap a bad one out if one fails in the RAID array. Buy an extra RAID card if you think it may fail. Buy an extra power supply. Buy an extra stick of RAM incase one goes bad. Buy an extra motherboard incase it dies. Buy an entire extra computer exactly the same, that gets backed up to hourly, incase the first one blows up or flood/fire/earthquake.

Fact is, there is going to be *some* down time at some point whether it's minutes or hours. If your boss is anal enough to fire you over 1 minute down time then that's not someone anyone should be working for. You can only prepare for so much before it gets insanely expensive.
 
nk215 said:
Dave;

I never said I put the data onto the boot drive. If/When I go with the X25-E then it would be two X25-E drives. I didn't state that clearly in the above post.

I'll take your advice and wait for the new DC 3700. Thank you.

I think about RAID 1 a lot lately, it comes down to this:

1) I don't trust the built-in raid - had bad experience with it before
2) I can get a hardware raid card. That's not the problem but I face similar question: what's the chance of the card itself fail vs. the X25-E SSD fail?

Up-time is only as good as the least reliable part. If the raid card is not as reliable as the X25-E then what's the point of using it at all?
Click to expand...

The RAID cards, traditionally, have allowed arrays to be interchangeable (and recoverable) where motherboard RAID didn't. My understanding is that recent Intel ICHs are better about that though.

Statistically, drives fail a lot more often than motherboards or RAID cards, and RAID is a relatively affordable way to mitigate the downtime and data loss potential. (It's all gambling with statistics anyway.)

For really mission critical stuff, having a hot-spare system running as Glacier mentioned is totally a thing people do - it's just ludicrously expensive. I don't run a spare everything, but I'm too old, have too much disposable income, and have buried too many hard drives to build a server w/o RAID anymore.
 
I, in fact, do have a spare in case the first computer goes down because of hardware issue (I'll just pull the data HD, put it in the spare and up it goes). I just never have 2 computers online at the same time because they can't have the same address.

Your post gave me this thought: I can just have both computer on, and have the data sync in real time from the main to the backup (drop box or something similar). When the main computer goes down, I'll just change the IP address of the backup comp to the main's IP address and everything should be back to normal.

In the past 10 years, I never have to to use the backup computer (knock on wood). The thought of having both running never occurs to me.
 
With a 128GB Samsung or Crucial @ 100 dollars you can't go wrong. Neither hard drive or SSD should die prematurely if you take care of it. With hard drives defrag, and always check for bad sector,, if you have them, you can get rid of them with HDD Regenerator and you have 0 bad sectors again.. Also make sure its screwed on tight and unmovable. Also make sure its temp is not more then 34c in its life. With SSD ,, always gotta do firmware updates and use CrystalDisk. Go this route for speed and reliability and no noise. gl
 
Last edited:
tweakboy said:
with SSD ,, always gotta do firmware updates and use CrystalDisk. Go this route for speed and reliability and no noise. gl
Click to expand...

No you do not have to ALWAYS do FW updates. Unless the manufacturer states it is a 'critical update' (as was the case with the Crucial M4 0009 FW) you really do NOT need to do FW updates. Running crystal disk only checks your performance - which you DO NOT need to do unless you a benchmark junkie or want to prematurely kill your SSD.

Please stop giving misinformation.......
 
If your job depends on it, I wouldn't go with consumer drives. Enterprise drives cost a magnitude more for a reason (e.g., much more validation + higher quality NAND chips).
 
I would recommend a reliable set of Samsung or Intel SSD's. I really think RAID 10 is best for allowing 1-2 drive failures as well as good performance.

If you are looking for that 99.99% uptime sort of goal then you need HA (high availability). Look into VMWare HA, if one server goes down you can switch to backup hardware.

Also backup everything as often as possible because everything made by man will eventually fail.
 
Some may call me out on this, but technologies like RAM or CPUs (if not abused) either fail early during a burn-in period, or tend to last. I would say that SSDs fall into that category, although there is apparently a finite number of writes that can be made to SSD cells. And to be honest, I'm still not much familiar with how SSDs work . . . but . . .

But. The MTBF for an SLC SSD is supposed to be something like 200 years. The MTBF for MLC SSDs is supposed to be 10 years, or I've seen it stated in millions of hours.

While you could corrupt OS files on an SSD just as with an HDD, I'd feel somewhat less worried about SSD failure on a well-maintained system. That is, the odds of failure seem lower.
 
I will say SSD is reliable than Mechanical HDD. You must use it on other way for all the factors. There are SSD wear indicators / tools available or come with it.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top