Jump to content

System Testing and Benchmarks


NoelC

Recommended Posts

Well, I don't have an analysis suitable for a Tom's Hardware article at hand, but the SSDs I bought have a 2 million hour MTBF.

 

Even enterprise-class HDDs at the time (e.g., Western Digital RE4, of which I have two) were only pushing 1 million hour MTBF, with consumer drives presumably less.  Let's say a half million hours in round numbers.

 

I roughly figured RAID 0 with any single drive failure causing loss of the whole array:  2 million hours MTBF for any one drive / 4 drives in the array = a half million hours MTBF for the array.  Roughly the same reliability as a typical HDD.

 

And it's just common sense.  A solid state device seems less likely to fail than an electromechanical one.  Beyond that, consider that HDDs use more power and get a good bit hotter than SSDs.  Heat kills electronics.  My fans almost never increase above idle.

 

Keeping in mind RAID 0 spreads the write load across several drives, one has only to make the array out of sufficiently large SSDs so that NAND wear is not a practical issue.   I figured mine would last 10 years at the write load I use before NAND wear becomes an issue - and I use my system pretty heavily.

 

-Noel

Link to comment
Share on other sites


I asked because I initially understood you previously quoted sentence backwards (not understanding at first sight that you meant specifically a RAID 0 setup ), all in all (partially) my bad  :blushing: .

If a HD MTBF=1,000,000, and a SSD MTBF=2,000,000, a 4 device RAID 0 of SSD's has approximately 1/4, i.e. 500,000, i.e. 1/2 of the single drive, now that you cleared how you half the MTBF of the hard disk before everything is clear.

 

As a side note there are however IMHO few things as inaccurate (or as wrongly perceived, you choose) as the rated MTBF, I have found flippism as reliable as SMART technology, and I would say that roulette playing should be almost as reliable as trusting the MTBF declared by the manufacturer :w00t::ph34r:.

http://www.dailytech.com/Study+Hard+Drive+MTBF+Ratings+Highly+Exaggerated/article6404.htm

 

I personally consider MTBF one of the many almost meaningless metrics around (i.e. only useful - maybe - for comparisons but not in any way a method to appreciate a "real life" measure).  

 

 

jaclaz

Link to comment
Share on other sites

Well, a correctly derived MTBF is meaningful - but maybe less so in the consumer world since Marketing seems to drive everything now and they're comfortable with outright lying.  And who's going to actually test a whole bunch of parts to see if a 2 million hour MTBF is accurate?

 

I personally believe consumer HDDs have MUCH less reliability (probably much less than half the true MTBF) as compared to enterprise class HDDs.  That's a lifetime of experience talking.  When I started buying enterprise class drives I stopped having data loss.  And there are some indications that in this day and age of "buying on price alone" things have gotten worse.

 

The brand of SSD drives I bought (OCZ) is by a company that makes enterprise class hardware as well, and the fruits of my research (which went a bit beyond reading the datasheets) implied their hardware is quite reliable (not to mention well-supported).

 

Bottom line is I've gotten 2 years and 2 months of reliable service out of 4 OCZ Vertex 3 drives under reasonably intense daily use.  I'd buy the brand again, on the assumption that their current hardware is as well-designed and manufactured. 

 

That being said, modern Samsung models sport slightly better numbers and since Samsung manufactures flash themselves, slightly better prices.  Specifications always tempt...

 

-Noel

Edited by NoelC
Link to comment
Share on other sites

Sure :), and I do believe in your results and do not even think of doubting about your choices, which surely were taken after some good research and thinking :yes:, I had however the impression (possibly false) that in that post you attempted to rationalize :w00t: your choices using as base the MTBF (which is as said IMHO in itself largely a meaningless metric).

 

I.e. it seems to me like you did most probably the right choice, but you did that because of your lifetime experience and (inborn or acquired) common sense and not because of the published MTBF's. or these played however a minor role in your choices.

 

jaclaz

Link to comment
Share on other sites

Not exclusively on MTBF, no.  I try to use all the knowledge and wisdom I have accumulated to make choices.  And I try to share that knowledge when I can.  That's one of the reasons I love forums like this one.  It's also one of the reasons I started this thread - because I'm always looking for ways to improve that knowledge.  Regular, ongoing benchmarking is one way I keep in touch with how things are working.  Simply put, I watch for problems / trends using objective data.

 

Regarding the use of published MTBF...  It's one input, and like it or not bigger numbers are generally better.  They tend to indicate how confident the manufacturer is in their products, and are arguably more meaningful with higher-end products, which in an engineering company I tend to be involved with.

 

This round of questioning started with you asking me for specifics about my comment on an SSD array being as reliable as an HDD.  I figured quoting numbers as one way to justify the comment made sense.  But you're right, it's VERY difficult to compare apples to apples when considering different technologies.

 

-Noel

Link to comment
Share on other sites

Speaking of the devil, wow...

 

Samsung 840 Pro 512 GB at Newegg is $379.99. 

 

OCZ Vector VTR150 480 GB is $359.99.

 

I love the way prices drop.

 

-Noel

Link to comment
Share on other sites

Speaking of the devil, wow...

 

Samsung 840 Pro 512 GB at Newegg is $379.99. 

 

OCZ Vector VTR150 480 GB is $359.99.

 

I love the way prices drop.

 

-Noel

I picked up the 512 840 EVO from Newegg @ $264.99  http://www.newegg.com/Product/Product.aspx?Item=N82E16820147249

http://www.samsung.com/us/system/consumer/product/mz/7t/e2/mz7te250bw/Sam_1303_SSD-840-EVO-Spec-Sheet_v9.pdf

http://www.samsung.com/us/system/consumer/product/mz/7t/e2/mz7te250bw/Sam_1303_SSD-840-EVO-Spec-Sheet_v9.pdf

 

Happy Camper 

~DP 

Link to comment
Share on other sites

I believe I have heard - and don't quote me on this - that the Pro may be better than the Evo for use in RAID operation.  However, most agree that all the Samsung models are smokin' fast.

 

These are just made out of chips.  Once production ramps up prices will fall more.  Good times.

 

-Noel

Link to comment
Share on other sites

Yep Fast as you know what, boot-up on a older 2nd gen I7 is under 10 secs, Firmly believe that the most significant difference if you could only make one improvement would be going with a SSD!

 

~DP 

Link to comment
Share on other sites

Absolutely true.  Only thing better (not counting exotic stuff) is more SSDs, but even then (as has been discussed) the gain won't be nearly as great as that first one gives over an HDD.

 

And the computer is blessedly quiet.  When I (rarely) spin up the old hard drives I'm surprised at how irritating that little tickety tick seek noise is.

 

Once you've used an SSD-equipped system you'll never want to use anything else.

 

-Noel

Link to comment
Share on other sites

I also own OCZ SSDs and they keep working fine, but facts are tough:

 

Return rates April-October 2013

 

SSD:

- Samsung 0,54% (contre 0,28%)

- Sandisk 0,70% (N/A)

- Kingston 0,72% (contre 1,00%)

- Intel 0,90% (contre 0,63%)

- Corsair 0,91% (contre 1,88%)

- Crucial 1,08% (contre 2,26%)

- OCZ 5,66% (contre 2,27%)

 

 

HDD:

- Seagate 0,86% (contre 0,95%)

- Toshiba 1,02% (contre 1,54%)

- Hitachi 1,08% (contre 1,16%)

- Western 1,13% (contre 1,19%)

 

http://www.hardware.fr/articles/920-1/taux-retour-composants-10.html

Now that Toshiba has acquired OCZ after its bankrupcy I expect reliability will increase.

 

 

 

On price drop, the 120GB Vertex 2 in the first page cost me about US$200 in late 2010.

 

Today for the same price we can get a 512GB Crucial MX100 of four times the capacity @ double the speed. :thumbup

I don't even contemplate booting a system from a spinner anymore.

Link to comment
Share on other sites

When OCZ first came out with the sandforce controller, I had bsod's like crazy  and they would never take ownership of the problem but believe Samsung does there own controller now in house and never had one hiccup (guess i like sams)

 

~DP

Link to comment
Share on other sites

 

Today for the same price we can get a 512GB Crucial MX100 of four times the capacity @ double the speed. :thumbup

 

 

I wait a few weeks and check if users have issue or not. If not, this will replace my Samsung 830 256Gb SSD. The 256GB are too small now.

Link to comment
Share on other sites

 

I wait a few weeks and check if users have issue or not. If not, this will replace my Samsung 830 256Gb SSD. The 256GB are too small now.

 

Would you mind if I add as a corollary that recent OS's are too d@mn big now? :unsure:

 

jaclaz

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...