So, I've a notebook with 1x sata 6gbps and I want to up an SSD on it.
I already have a crucial M4 256gb but I need more storage. I need 500gb.
I'll have to use the M4 256gb on other setup so I'll always have to buy 500gb of SSD/s.
My doubts are if I would benefit from using a 256gb plus another 256gb on the ODD. I don't use the ODD at all (in fact I don't remember using it since I've the notebook, not even for OS installation).
So, should I use 2x256Gb using software RAID or just get a 512Gb SSD?
-
Get a single SSD (512GB). AFAIK trim hasn't been supported (yet) in any raid configuration leading to performance degradation over time.
-- -
Lot's of things to think about - increased read/write times, but also write amplification issues due to lack of TRIM, greatly increasing your chances of catastrophic data loss ( but everyone is backing things up, right? ).
You may want to look at this thread for more info - http://forum.notebookreview.com/har...-0-sata-2-x2-ssds-vs-single-ssd-sata-3-a.html -
If you just need storage space, why not consider a 1TB HDD?
-
Unless you have >256GB of data that your applications absolutely have to enjoy low latenty and high throughput (which I highly doubt):
-
Then again, if he's considering a 500+GB SSD, then I assume money is no object.
-
Money might not be an object, but that doesn't mean you have to be stupid with it
.
-
So I've decided to get a single SSD because of the sata III interface. RAID by software using the lower bandwidth interface of ODD could actually lower the performance.
A multiple SSD setup only has one advantage: speed.
I would like this speed but it seams I can't improve it. things could be different if I had 2x sata III slots but that's not the case.
I disagree that the risk of data lost is increased using raid. If 1 drive is going to fail then it fails in raid or not.
But yes it has many disadvantages, some major ones some minor ones. About TRIM I think it depends on the drive. There's actually SSD's that manufacturers recommend to turn TRIM off to avoid messing with the drive garbage collection system. Don't known if would be affected by RAID but I don't think so.
Also there are obvious disadvantages, unable to use the ODD (though I don't use it), more weight (SSD drives weight like 100g... though it can be increased if used with a 1.8" to 2.5" adapter or an ODD adapter), double of SSD power consumption.
Conclusion, if I had 2x sata III I would search information to make my choice. But I don't so a single SSD solution ftw. -
Sounds like that will work for you. A couple of comments/corrections to your post.
-
link: To TRIM or not to TRIM (OWC has the answer) | Other World Computing Blog -
That article is for an OWC SSDs and on a Mac. You are (were) on an Crucial M4, and presumably not on OSX.
So, while it is interesting for OWC SSDs, it doesn't apply in your case (again assuming a 2nd M4), but I do wonder how this applies to other SandForce based SSDS as well as when GC will run in Win 7 vs. Mac OSX.
In addition there are two quotes from a reference on that article that are also interesting:
Thanks for the link. -
jclausius said: ↑But since OS X Lion and/or Trim Enabler now allow TRIM, does that mean TRIM will keep an SSD at its original performance level even on OS X Lion - similar to PC users?Click to expand...
jclausius said: ↑Thanks for the link.Click to expand... -
__-_-_-__ said: ↑I think it refers that only some SSD's support TRIM with RAID. those must be 'rated for it' while others don't.Click to expand...
Oh, well. No real way to tell what they were thinking. -
__-_-_-__ said: ↑I disagree that the risk of data lost is increased using raid. If 1 drive is going to fail then it fails in raid or not.Click to expand...
Lets say 1000 drives are produced and one of them in that 1000 is going to fail early then a person buying 4 drives to use from that batch is twice as likely to get the duff drive than someone buying 2 to use in RAID.
Distribution can also be taken a step further. If 1 person in 999 buys 2 of those drives for RAID and the other 998 people buy 1 drive each to use as standalone then the chances that the RAID user gets the duff drive is 0.2%. The chances that a standalone drive user gets the duff drive is 99.8%. In other words it is ~500 times more likely one of the users of 1 drive will get the duff drive.
Those who say you double the risk can make it sound like there is going to be imminent failure whereas IMO if your unlucky enough to get a duff drive then just as a non RAID user might replace the disk and re-install / restore so too might a RAID user. The only drawback is the array life before failure will be that of the worst drive but if the worst drive is going to last 5 years or more then maybe it's not such a big deal. -
Those who say you double the risk can make it sound like there is going to be imminent failureClick to expand...
For those who don't understand, you need to go back to chance and probability theory. In the case where ONE out of a number N fails, creates a dependency of that failure. So, just by adding N number of drives, the *mathematical* fail rate of the entire RAID volume to fail has to goes up.
So, let's assume the chance of a something going wrong is .0001 (.01%). Pretty small right. So ... successful RAID = 1.000 - .0001 = .9999. A 99.99% chance that things are just fine. Seems pretty good.
Now let's go w/ a 3 drive array. Remember if any one of them dies the entire RAID fails. Due to this dependency, each drive's success rate must be taken into account. In this case, .9999 * .9999 * .9999 will determine the success rate. Or, 0.0.999700029999 or 99.97%. A 4 drive array would be 0.9996000599960001 or 99.96%.
To simplify, the formula for calculating RAID-0 *success* is (1.00 - CHANCE_OF_FAILURE)^N. Where N is the number of drives.
So, you can see, in this case, 99.99% success vs. 99.97% success. Both are pretty good chances, but the 3 drive array is just a tad worse (.01 vs. .03) It is up to the decision maker if they can live with this chance.
So, when someone says the risk of data lost is *NOT* increased as you add more drives to a RAID-0 volume, they are mathematically mistaken, as this is fact that cannot be argued. But the chances may be so miniscule, it may not matter to the RAID-0 user.
Hope this explains things. -
raid it's not just 1^N drives. the raid setup itself increases greatly the risk of something go wrong. this is, having 2 separate drives working is more secure then having a raid array where 2 drivers work combined, specially in SSD's.
Also chances of something go wrong is far greater then 0.01%. often +2%.
failure rates:
Disques durs & SSD - Les taux de retour des composants (3) - HardWare.fr -
The .0001 was an arbitrary number. Try 2% in the calculation above. You will see 98% success rate w/ one drive and just around 96% rate for a two drive RAID-0 array -> (.98) * (.98) = .9604 OR (1.00 - .02)^2 = .9604
I'm not sure I understand your second through fourth sentences, but to summarize, RAID-0 the risk of losing the entire volume *increases* with each added drive, but on a different (and off-topic) note, the risk of losing the volume *decreases* for each added drive with RAID-1. -
jclausius said: ↑To simplify, the formula for calculating RAID-0 failure is (1.00 - CHANCE_OF_FAILURE)^N. Where N is the number of drives.Click to expand...
jclausius said: ↑It is up to the decision maker if they can live with this chance.Click to expand...
__-_-_-__ said: ↑the raid setup itself increases greatly the risk of something go wrong. this is, having 2 separate drives working is more secure then having a raid array where 2 drivers work combined, specially in SSD's.Click to expand...
__-_-_-__ said: ↑Also chances of something go wrong is far greater then 0.01%. often +2%.Click to expand... -
Dufus said: ↑Shouldn't that be 1-((1.00 - CHANCE_OF_FAILURE)^N) or did you mean the chance of not failing?Click to expand...
to RAID or not to RAID?
Discussion in 'Hardware Components and Aftermarket Upgrades' started by __-_-_-__, Dec 22, 2011.