Since an SSD and an HDD are essentially two drives, when combined doesn't the probability of the unit failing go up? Any idea how much?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Drive Failed, considering hybrid - what to look out for?
- Thread starter BradHatter
- Start date
- Sort by reaction score
I don't know, but if the solid state part of an SSHD fails you should not expect the drive to fail. Simply expect its performance to drop to what's possible with the hard disk part.
Not really. The SSD portion of a hybrid acts more like a cache, not something like a Fusion drive, and it's usually part of the drive controller. A failure of the SSD section would likely be a failure of the controller itself. If the SSD failed itself it's hard to tell what a given controller on a given unit would do.
Is there any way to tell just exactly what is stored where on a hybrid (i.e. what's in the SSD cache?)
I've been reading some other posts on dual SSD/HD manual configurations and it sounds to me like the best performance gain might be to manually configure what goes where. For example, someone could put the SSD in, treat it like it was a regular drive, install the OS on that, make the HDD active, move whatever they wanted onto that, then use symbolic links to link back into the OS.
You would know what was on which unit at what time and you could tweak it for performance as needed. No guessing at all.
I've been reading some other posts on dual SSD/HD manual configurations and it sounds to me like the best performance gain might be to manually configure what goes where. For example, someone could put the SSD in, treat it like it was a regular drive, install the OS on that, make the HDD active, move whatever they wanted onto that, then use symbolic links to link back into the OS.
You would know what was on which unit at what time and you could tweak it for performance as needed. No guessing at all.
No. There's no (easy) way to tell what's in a cache at a given time. Fusion drives would have a similar problem because the software would be bouncing components, as in files or sections of files from the SSD to HDD and back. This is one of the reasons some people are complaining about them.
Since you use Scannerz, you could, assuming you're familiar with command line descriptions of files and locations, use the FSE or FSE-Lite product included with it to track all the heavy file system hitters, make a symbolic link from them to the SSD and put the heavy work load there. You could also control backups with Time Machine by adding components to the backup, so you would know what is where at what time and be able to back it up. Assuming the SSD would host the core operating system, you could probably put a small volume with just the OS on the hard drive in the event something went wrong with the SSD.
I noticed, and I think it was in this thread, that a former Fusion drive user complained that using a video conversion application took a fraction of the time on an SSD compared to a Fusion. This would be a good test. Such an application often writes tons of temporary data to a temporary location and it's often in the users Library directory. If that directory were linked to the SSD instead, then likely a noticeable performance increase would be experienced, but the core file would probably still end up on the hard drive.
That wouldn't be as fast as a pure SSD, but if tweaked and managed properly, I bet it could come close to approaching it speed wise.
Manually configuring your own "hybrid" storage on a combined SSD and HDD solution may really be the best way to go. You know what's where, and you can control backups as you see fit.
What I don't like about Fusion drives are that you never know what file, or for that matter, portion of a file is on which drive at which time, and Apple has mad the whole operation "Top Secret" as usual. OK, if you want to keep it a secret, fine, just don't expect everyone to adopt it.
The SSHD, I assume, caches the stuff that's most often hit, but this would, if my assumption is right, likely toss a lot of stuff onto the hard drive. Once again, no one knows the algorithms, and they probably vary from manufacturer to manufacturer, and firmware release to firmware release.
What I don't like about Fusion drives are that you never know what file, or for that matter, portion of a file is on which drive at which time, and Apple has mad the whole operation "Top Secret" as usual. OK, if you want to keep it a secret, fine, just don't expect everyone to adopt it.
The SSHD, I assume, caches the stuff that's most often hit, but this would, if my assumption is right, likely toss a lot of stuff onto the hard drive. Once again, no one knows the algorithms, and they probably vary from manufacturer to manufacturer, and firmware release to firmware release.
I came across this yesterday:
Original article link:
http://www.infoworld.com/article/28...n-digital-hgst-most-reliable-hard-drives.html
I assume problems with a "regular" HD might show up in hybrids as well. Now I'll probably find out that InfoWorld just got a big contract with WD/Hitachi not that I'm cynical or anything.
Original article link:
http://www.infoworld.com/article/28...n-digital-hgst-most-reliable-hard-drives.html
I assume problems with a "regular" HD might show up in hybrids as well. Now I'll probably find out that InfoWorld just got a big contract with WD/Hitachi not that I'm cynical or anything.
How does that drive do with something that's drive intensive, like a video conversion. Video conversions typically write tons of temp data to the drive and it would be interesting to see if it caches that sort of thing on the SSD portion of the drive or the actual hard drive. Needless to say, the SSD part would be faster.
SSDs are just too expensive for mass amounts of data and there are lots of people with just that - mass amounts of data. It's also a newer trend because the price of SSDs themselves used to be so expensive but now a small SSD can be put in with an HDD and be fairly cost effective.
I think this trend will grow, not disappear or decline.
I don't think HDDs are going to disappear any time soon. Most HDDs last 3-5 years or longer and they can store TB of data for tens of dollars. SSDs, although clearly improving, still seem to have some glitches and for anyone with any real amount of data, they're just too expensive.
I came across this yesterday:
Image
Original article link:
http://www.infoworld.com/article/28...n-digital-hgst-most-reliable-hard-drives.html
I assume problems with a "regular" HD might show up in hybrids as well. Now I'll probably find out that InfoWorld just got a big contract with WD/Hitachi not that I'm cynical or anything.
That's their desktop drive line. The seagate laptop sized drives seem better.
----------
Yeah, basically there's no reason to use an SSD to store movies and audio. You don't need 500 MB/s read speeds for those kinds of files. Also, the server market buys hard drives by the thousands for cost per gigabyte and scaling reasons.
I surely hope for Seagate's sake that data isn't true. Personally I'm a little suspect.
Regarding the storing movies and audio, well put. We used to access these with CDROMs and DVDs, which of course make hard drives look like speedsters.
I think someone disputed the Seagate data in another thread. I don't have the link, but I don't think I'd give that report much validity.
However, they do seem to be having a fair amount of more problems than the other vendors quality wise. I hope they resolve their problems. I always used to like Seagate.
However, they do seem to be having a fair amount of more problems than the other vendors quality wise. I hope they resolve their problems. I always used to like Seagate.
There are a number of critiques about that article on the web, like this:
http://www.theregister.co.uk/2014/02/17/backblaze_how_not_to_evaluate_disk_reliability/
Some of the Seagate's were apparently quite old.
The trouble with people who aren't prepared to read linked information
From that February 2014 article How NOT to evaluate hard disk reliability: Backblaze vs world+dog:
From Henry Newman's January 2014 article:
That's really not a great job of digging into the guts of the Backblaze analysis. It seems that Newman did not follow the links offered by Backblaze.
From the October 2013 post by Backblaze:
So. Backblaze mentioned "burn in" six times in a single article, then the 'great' Henry Newman asked "does Backblaze have a burn in period?".
Am I missing something?
From that February 2014 article How NOT to evaluate hard disk reliability: Backblaze vs world+dog:
"
Henry Newman does a great job of digging into the guts of the Backblaze analysis and pointing out the shortcomings in their approach.
"
From Henry Newman's January 2014 article:
"
This is never discussed in terms of: does Backblaze have a burn in period?
"
That's really not a great job of digging into the guts of the Backblaze analysis. It seems that Newman did not follow the links offered by Backblaze.
From the October 2013 post by Backblaze:
"
initial burn in of the drives as they are built into RAID arrays. This requires that every block on every drive is read and checked. For a 45 drive Storage Pod this step typically takes anywhere from 2 to 5 days depending on the drives being tested. We use the day-per-terabyte rule of thumb to estimate the time needed. For example, if a Storage Pod has 4TB drives we expect it to take 4 days to burn in.
When the burn in process terminates we take another SMART snapshot of each of the drives. To pass the burn in phase, the RAID arrays have to be successfully built and synced, and all of the drives have to pass their SMART review that includes a comparison to the first SMART snapshot taken earlier. If anything has failed along the way, the failed component is identified and replaced and the burn in phase is restarted from the beginning. We start from the beginning because you are never sure what could have been disturbed in fixing the failure and more importantly spending a few extra hours now to rerun the entire test has proven to save us countless hours later. "
When the burn in process terminates we take another SMART snapshot of each of the drives. To pass the burn in phase, the RAID arrays have to be successfully built and synced, and all of the drives have to pass their SMART review that includes a comparison to the first SMART snapshot taken earlier. If anything has failed along the way, the failed component is identified and replaced and the burn in phase is restarted from the beginning. We start from the beginning because you are never sure what could have been disturbed in fixing the failure and more importantly spending a few extra hours now to rerun the entire test has proven to save us countless hours later. "
So. Backblaze mentioned "burn in" six times in a single article, then the 'great' Henry Newman asked "does Backblaze have a burn in period?".
Am I missing something?
I said it before and I'll say it again: He who lives by the third party review dies by the third party review.
Tha Back Blaze article has been criticized by more than one source, with one arguing it was a marketing stunt by the company to draw attention to themselves. The failure rate of Seagate's is abnormally high, and I think so abnormally high it draws attention and scrutiny to the article itself.
Tha Back Blaze article has been criticized by more than one source, with one arguing it was a marketing stunt by the company to draw attention to themselves. The failure rate of Seagate's is abnormally high, and I think so abnormally high it draws attention and scrutiny to the article itself.
There are a lot of critiques of the Backblaze article. Apparently one of the drive's they're basing the claims on was a Seagate drive that Seagate had problems with and may have warrantied.
A real manufacturers sample wouldn't be based on a single or few models it would be based on all product lines.
That "article" also seems to be missing a few brands. Toshiba and Fujitsu come to mind. How do you conclude that you've done decent tests on drive brands and determined who's the best when a few major brands aren't even included.
I did notice, however, that the Backblaze logo in on the graph (free advertising!)
I did notice, however, that the Backblaze logo in on the graph (free advertising!)
Backblaze: further reading
DT01ACA300 is mentioned by Backblaze in at least four of their blog posts.
More generally, https://duckduckgo.com/?q=Toshiba site:backblaze.com shows Toshiba mentioned in a few other places, most notably What Hard Drive Should I Buy? (2014-01-21) there's the original graph to which the opening poster drew attention. Backblaze explained:
Also, Reliability Data Set For 41,000 Hard Drives Now Open Source (2015-02-04) as expected, the set includes some Toshiba data.
No trace of Fujitsu in that data set. I guess that for the reasons given above and below, no Fujitsu drive went on the buy list.
Validations, quotes for bulk purchase
From What is the Best Hard Drive? (2015-01-21):
Failures and impending failures of drives
From today's https://www.backblaze.com/blog/backblaze-blog-no-comment/#comment-1945367580 :
DT01ACA300 is mentioned by Backblaze in at least four of their blog posts.
More generally, https://duckduckgo.com/?q=Toshiba site:backblaze.com shows Toshiba mentioned in a few other places, most notably What Hard Drive Should I Buy? (2014-01-21) there's the original graph to which the opening poster drew attention. Backblaze explained:
"
We dont have enough Toshiba or Samsung drives for good statistical results.
Why do we have the drives we have? Basically, we buy the least expensive drives that will work. When a new drive comes on the market that looks like it would work, and the price is good, we test a pod full and see how they perform. The new drives go through initial setup tests, a stress test, and then a couple weeks in production. (A couple of weeks is enough to fill the pod with data.) If things still look good, that drive goes on the buy list. When the price is right, we buy it.
We are willing to spend a little bit more on drives that are reliable, because it costs money to replace a drive. We are not willing to spend a lot more, though. "
Why do we have the drives we have? Basically, we buy the least expensive drives that will work. When a new drive comes on the market that looks like it would work, and the price is good, we test a pod full and see how they perform. The new drives go through initial setup tests, a stress test, and then a couple weeks in production. (A couple of weeks is enough to fill the pod with data.) If things still look good, that drive goes on the buy list. When the price is right, we buy it.
We are willing to spend a little bit more on drives that are reliable, because it costs money to replace a drive. We are not willing to spend a lot more, though. "
Also, Reliability Data Set For 41,000 Hard Drives Now Open Source (2015-02-04) as expected, the set includes some Toshiba data.
No trace of Fujitsu in that data set. I guess that for the reasons given above and below, no Fujitsu drive went on the buy list.
Validations, quotes for bulk purchase
From What is the Best Hard Drive? (2015-01-21):
"
only one Storage Pod of Western Digital 4 TB drives. Why? The reason is simple: price. We purchase drives through various channel partners for each manufacturer. Well put out an RFQ (Request for Quote) for say 2,000 4 TB drives, and list the brands and models we have validated for use in our Storage Pods. Over the course of the last year, Western Digital drives were often not quoted and when they were, they were never the lowest price. Generally the WD drives were $15-$20 more per drive. Thats too much of a premium to pay when the Seagate and HGST drives are performing so well.
"
Failures and impending failures of drives
From today's https://www.backblaze.com/blog/backblaze-blog-no-comment/#comment-1945367580 :
"
Parts of https://www.backblaze.com/blog/hard-drive-smart-stats/ (2014-11-12) and of https://www.backblaze.com/blog/vault-cloud-storage-architecture/ (2015-03-11) are relevant to that commentary.
Still, I'd like to see a more cohesive blog post or short series of posts about, for example:
Still, I'd like to see a more cohesive blog post or short series of posts about, for example:
- whether Backblaze would treat a pattern of shard checksum mismatches as an indication of failure of a drive, or impending failure of a drive with no comparable indication from the S.M.A.R.T. metrics that are currently preferred by Backblaze.
Am I missing something here?
" only one Storage Pod of Western Digital 4 TB drives. Why? The reason is simple: price. We purchase drives through various channel partners for each manufacturer. Well put out an RFQ (Request for Quote) for say 2,000 4 TB drives, and list the brands and models we have validated for use in our Storage Pods. Over the course of the last year, Western Digital drives were often not quoted and when they were, they were never the lowest price. Generally the WD drives were $15-$20 more per drive. Thats too much of a premium to pay when the Seagate and HGST drives are performing so well. "
Didn't they just give Seagate what I'd call an "F" and then they make the quote "Thats too much of a premium to pay when the Seagate and HGST drives are performing so well. "
" only one Storage Pod of Western Digital 4 TB drives. Why? The reason is simple: price. We purchase drives through various channel partners for each manufacturer. Well put out an RFQ (Request for Quote) for say 2,000 4 TB drives, and list the brands and models we have validated for use in our Storage Pods. Over the course of the last year, Western Digital drives were often not quoted and when they were, they were never the lowest price. Generally the WD drives were $15-$20 more per drive. Thats too much of a premium to pay when the Seagate and HGST drives are performing so well. "
Didn't they just give Seagate what I'd call an "F" and then they make the quote "Thats too much of a premium to pay when the Seagate and HGST drives are performing so well. "
Getting sort-of back on topic here, has anyone heard of using an SSD and an HDD in combo and some software that would off load the less used stuff to the HDD and put the higher speed stuff on the SSD.
I know that what I'm describing sounds a lot like a Fusion Drive, but this isn't dynamically managed. These would be actual partitions with the physical partitions on the HDD linking into the SSD to make it look like they were part of the SSD.
I know that what I'm describing sounds a lot like a Fusion Drive, but this isn't dynamically managed. These would be actual partitions with the physical partitions on the HDD linking into the SSD to make it look like they were part of the SSD.