Toshiba PX04S 12Gb/s SAS Enterprise SSD Review

[Guest]

The 12GB/s SAS Toshiba PX04 barrels onto the data center scene with PCIe-like speed and the inherent advantages of SAS.

Toshiba PX04S 12Gb/s SAS Enterprise SSD Review : Read more

 

[dgingeri]

Given my experiences with Toshiba enterprise level HDDs, I wouldn't trust this thing with my data. I had over 1200 Toshiba 146GB 15k drives across my lab in 2010. Every single one of them died before they hit 3 years old. Fortunately, we didn't lose data in any case, but it was almost a full time job just swapping out drives and calling Dell for replacements for the next day. I will not trust Toshiba enterprise level hardware again. I wouldn't accept such reliability problems in consumer level hardware, let alone enterprise level.

 

[jeff_sloan]

dgingeri, Seriously? Over 1200 hard drives failed before 3 years?
You make it sound like that was ALL your drives in your lab, but I find it hard to believe there was a 100% failure rate, so how many Toshiba drives did you have total? How many of all brands together?

 

[captaincharisma]

@dgingeri you do realize they are talking about solid state drives and not mechanical hard drives right?

considering you mentioned drives from 2010 i don't think toshiba made any SSD drives let alone have any hard driver under there name at that time

 

[nadr6969]

wow

 

[YourLocal]

On a different note on the consumer side the only SSD that has ever died on me was a OCZ Vector 150. Luckily I bought it at newegg and was in the 30 day return period. They refunded me a little under $300 for a 480GB drive. So not only did it fail, but it failed in about 2 weeks.

 

[dgingeri]

I am the admin of a test lab for server storage software development and for tape library testing. We have a LOT of hard drives. Just the performance department has 108 16 drive trays of FC storage. Over all three labs, I have approximately 600 FC trays. I also have around 400 SAS trays with 3.5" drives and about 150 SAS trays with 2.5" drives. Of the SAS trays, about 2/3 are just trays with SAS RAID controllers in the system, while the other 1/3 have their own RAID controllers or are attached to another tray with its own RAID controller. Many are partially filled, some are completely filled. I also have 90 Dell R910 systems with 13 or 15 2.5" SAS drives. I have configured and maintain every single one of those trays over the last 5 and a half years.

Now, addressing these drives in particular: When my company released a new product 4 years ago, based on the Dell R910 and two MD1220 trays, along with a lot of other storage, Dell took the lowest bidder on the drives we needed, 146GB 15k SAS. The winning bidder happened to be Toshiba, which is still using the Fujitsu name at the time. The specific drive was the mbe2147rc.

Our first systems for testing came in with 13 drives in the node and 17 drives in each MD1220 tray, three months before release to the public. Less than 3 months after release, we started having issues with drives failing. In some cases, they even failed in such a way that it spiked power back through the backplane and killed nearby drives and the backplane. Three months after that, we started getting complaints from our customers on the reliability of these drives. I was having to replace these drives 3-6 at a time on our test systems after less than a year. The reliability was HORRIBLE. We started getting drives with the Toshiba name on them instead of the Fujitsu name, but the same model number. They had the same issue. After the one year point, we told Dell to specifically use the Seagate Saavio series drive because of these reliability problems. We also started using the Seagate drives for all replacements. In the end, we did a full recall on all customer drives to replace them with Seagate drives. It cost us a pretty penny. They did not, however, issue that recall for internal drives, and many drives made it to me to use as spare parts. I got an entire pallet of those drives, 12 in each case and 12 cases, from our service department. I used them all before the three year mark on these drives. I have gone through since then to make sure we don't have any left. I recent received a system from our Irvine office that was loaded with them, and hadn't been turned on for over a year, and they are failing like crazy now. I built it as a VM host with those drives in a RAID 6 set, only to have the set die twice now because of too many failed drives at once.

My estimate of 1200 is probably low ball, and yes, that was every drive that I have had in this lab except for that most recent system. All my 146GB SAS drives are now Seagate Saavios, and I had the first one of those die this past weekend. In that time, 4 years, I've replaced a confirmed 3008 (188 cases of 16 each) Seagate Constellation ES 1TB and 2TB and Barracuda ES 1TB drives from our FC sets, but better than half of those are still in service. Yes, I do go through that many drives, about 1500 per year. My budget reflects that. I also spend about $3000 per year on FC cables and $1000 on cat 6 cables.

There was indeed a 100% failure rate within 3 years of those horrible drives, many taken out by the drives next to them. The failure was always the same, failure of the central bearing, either seizing the drive or leaving it sounding like a pepper grinder when running. Many of the ones that seized were the ones that sent the power spike through the backplane to kill other drives. That mostly happened in the central node, where the drives were stacked in 4s. Either the top or bottom drive would be left running, but due to the system config, running in mirrored pairs, would leave the system unable to run and had to have the software rebuilt.

Our newest lines use Seagate Saavio 2.5" and HGST 3.5" drives for the best in reliability, trying to salvage our company's reputation on that matter.

 

[mapesdhs]

Your experiences do of course sound terrible, but what has any of that got to do with this SSD product review? Also, at the time, did you not contact Toshiba directly?

I could equally well say that of all the conventional HDD brands I've dealt with, Fujitsu have been the most reliable, especially the MAX 15K series, but this means nothing when it comes to judging the merits of an SSD, consumer or otherwise. Over the years I've had plenty of bad Seagates, IBMs, etc., but one can't say an entirely different type of product should be avoided just because of what you've described.

If anything, I would be more interested to hear how Dell & Toshiba dealt with the failures from a support angle.

 

[dgingeri]

The reliability of the manufacturer of one item always has influence on other items they manufacture. If they don't put enough effort into the reliability of one enterprise product, they it is likely they won't put that same effort into other products. A company attitude like that is also not likely to change over a short period of time, either. It's a company wide or division attitude and management decisions on production quality, not an individual product design, that is to blame for most product reliability issues. Usually the attitude and management decisions of one division doesn't affect other divisions. In this case, they're both from the enterprise product division, so they'll likely have the same reliability problems.

 

[g00ey]

dgingeri; Your comments are appreciated but you do realise that you are talking about hard drives while the article is about SSDs? From a technical standpoint in terms of reliability and durability it is like comparing apples and oranges.

Also, you write that if a company don't put enough effort into the reliability of one enterprise product, they are not likely to put that same effort into other products. We all know that the Seagate Constellation drives are, or at least have been utter sheit, as you confirm yourself from your own experience. But yet you are praising the Seagate Saavio as the best in reliability. If the Saavio is as good as you claim then how much effort did Seagate put into it?

The same could apply for a "good" manufacturer; amidst their line of excellent products they may on occasion release a piece of turd every now and then.

 

[Uilleam]

I'd have to agree with dgingeri.

I've been a long time reader of Tom's since the 90's and have been a Computer Technician for 18 years.

Toshiba in my opinion is the worst Hard Drive made. Numerous times when Toshiba Laptops or other computers with Toshiba drives that would come into the shop and hear the customer articulate the symptoms I'd tell them their hard drive probably died. I'd say 80% of the time it was the drive. Usually the Click of Death.

Just because it's a Toshiba drive of any type I wouldn't trust it.

Other notable brands I wouldn't trust are:

Seagate (They do have firmware updates from time to time)
Maxtor
Samsung (They have good SSDs)
Fujitsu (Old Fujitsu used to be really good)

Brands I've had the best luck with are:

Western Digital
Hitachi
IBM

 

[mapesdhs]

Anyone can cherry-pick from personal experience, but that can't be a sensible rationale for a general conclusion.

I could equally say that the first standard WD I bought (2TB Red) was DOA, but does that mean I should conclude all WD drives are junk? Of course not, and indeed I've been using a VR 10K for some time in an older build (the first SATA I encountered which remotely impressed me compared to the SCSI/SAS tech I was used to).

Likewise IBM, I've dealt with lots of IBM SCSI drives, many of which were awful; should I extrapolate that to what they make today?

I bought various SAS drives in recent years, including Seagate, Hitachi and Fujitsu. The Seagates fared worst, with Hitachi not far behind. Only the Fujitsu survived long term, though it was a lesser 147GB 15K.

For older drives, ie. SCSI, Fujitsu has for me been far better than any other brand, which was a surprise given the early days rep Seagate & others had. However, I still bought a couple of hundred more Seagate 36GB 15K SCSIs recently (for upgrading old industrial-use SGIs) as sadly Fujitsu's equivalent SCAs do not support narrow/SE mode.

 

[nekatreven]

The reliability of the manufacturer of one item always has influence on other items they manufacture. If they don't put enough effort into the reliability of one enterprise product, they it is likely they won't put that same effort into other products.

This. Yes.

 

[ifIwasarichman]

Generally I work on this principle -
As we are but mere humans, it is inevitable that there will be problems.
It is not the problem that should define us, but how we deal with it.
So if Toshiba had a bad product and dealt with it's solution/remedy badly, there is no way I would deal with them again, unless there was a profound and obvious change for the better. Even then I would expect a heavy discount or free samples to test first.
This would be hugely beneficial to the companies reputation and bolster the confidence of potential customers. Remember, perception has a massive influence on peoples on purchasing decisions.
If they did not help dgingie with a appropriate solution to the problem, I can fully understand his opinion. Particularly in a business situation. I have had problems with Asus products and have had major troubles trying to get Asus to help (eg - not easy to contact, then slow to respond, 1x motherboard RMA took well over a year to get sorted, what was the point by then) - therefore I do not buy Asus products any more I don't care what they make, until they change and fix their customer service, they are a banned product.
Who knows if the team that created dgingie's problem have been changed due to cock-ups like he mentioned. But if the issues remain, after testing samples of a new product (and in dgingie's situation, Toshiba should supply them free of charge) it would reflect back onto their entire product range.

 

[mlee 2500]

Here's the thing dgingeri: Mapesdhs is right, and EVERY major drive vendor has released a lemon at some point....including Seagate (plenty of times)

In the 90's I deployed dozens of HP7000 series UNIX workstations, each with Seagate drives. All the drives came from the same bad manufacturing "lot", and over the course of several months nearly every one died on me and had to be replaced. Does that have any bearing on the Seagate drives you stand by? No, it doesn't.

That being said, I understand being averse to a vendor you had such a categorically bad experience with (though I doubt the extent and suspect you of embellishing), but extrapolating a quality assessment from a bad product to all other products a manufacturer puts out into perpetuity, would leave you with no vendors at all.

 

[mapesdhs]

mlee 2500, yes indeed, and in the case of SSDs, despite building up a strong reliability rep, both Intel and Samsung have messed up big time on occasion, but many seem to have forgotten (or they choose not to remember) Intel's 8MB bricked SSD issue, and atm I can't help feel that forum posters are being incredibly lenient towards Samsung re the 840 EVO old data issue, presumably because Samsung built up such a decent rep 'store' (if I can call it that) via their good older products like the 830 and original 840.

By contrast, the hate piled on OCZ long after any of the original-series Vertex complaints were relevant is just nuts, which is a shame because their later products like the Vertex4 and Vector were really good (heck, I have dozens of Vertex2Es and Vertex3s, all running fine). OCZ did make it all worse by a poor support response at the time, but that's not relevant anymore either. People need to move on.

I fully get that people are inclined to not touch a vendor again after a bad experience, but there's taking it to extremes. As you say, if one held this flag for all vendors that mess up, we'd end up not being able to buy anything from anyone.

My speciality is SGIs; I've seen numerous terrible Seagate, Fujitsu, Hitachi and IBM models, but likewise all these vendors have made good drives aswell.

Mind you, talking to a guy at a data recovery company way back, it seems that a lot of vendors get up to all sorts of tricks with disk manufacturing. The guy said one would hardly believe the things they find inside disks which fail sometimes, all sorts of dodgy hand-done fixes to issues so that drive can pass tests & be sold. A common trick is to disable one or more bad platters and sell the drive as a lesser capacity model, bit like CPUs with fewer cores via having some bad cores disabled (by disable I mean the platters still spin along with the others, but the controller board just can't see them, and the relevant heads are not present).

Ian.

 

[dgingeri]

While I consider Samsung to be the best for SSDs these days, I am fully aware that they have made some significant errors in their products. The 840 EVO, of which I own three, are a significant problem. I have two 120GB models and a 500GB model. I now only use the 120GB drives for temp files and swap file use on two machines. The 500GB drive is the host for my VM storage, but I am looking to replace it. Despite that, I have had more luck with Samsung than others. They haven't had significant problems with their 840 Pro or 850 Pro, so I'm fairly confident of those. It's just the cheaper lines that have issues.

It's the same with Intel. Their top of the line drives have been fairly well flawless. I have two X25-E drives that still work quite well as the boot/OS drives for a couple of my personal servers at home, and my employer has several hundred in use in my lab right now.

My problems with Toshiba aren't about one model drive. It's across the board issues, and what they do about it. While Intel has replaced many drives that had bricked due to firmware issues, and both Intel and Samsung have issued statements that they knew bout these issues and were working on fixing them, Toshiba has plugged their ears and shouted "LALALALALALALALALALA" trying to block out any cries of reliability problems with their products.

 

[mapesdhs]

And there you've hit the nail on the proverbial head, it's the support ethos which often matters more, though TBH Samsung has been really dragging it heals on the EVO issue, and they still haven't officially stated whether the standard 840 is affected aswell (user comments suggest it is). But sure, if Toshiba act like that across the board, then that's bad. No idea though personally, I don't have any of their storage products AFAIK.

Ian.

 

[dgingeri]

I don't believe the 840 EVO problem is solvable. I think it is an inherent problem with the TLC architecture. They'll probably stop using it soon because of this problem. Besides, all the problems TLC was intended to solve can now be solved with the 3D V-NAND: more storage in less space and cheaper production. Plus 3D V-NAND has much better endurance. The software solution they're using to "fix" the problem only increases the writes to the cells, reducing the usable lifetime of the drive.

 

[mapesdhs]

Yeah, bit of a catch22 "fix" isn't it?

I just don't understand why they refuse to make it clear whether the 840 is affected aswell. Users say it is,
but Samsung is keeping schtum.

Ian.

 

[ifIwasarichman]

mlee 2500, yes indeed, and in the case of SSDs, despite building up a strong reliability rep, both Intel and Samsung have messed up big time on occasion, but many seem to have forgotten (or they choose not to remember) Intel's 8MB bricked SSD issue, and atm I can't help feel that forum posters are being incredibly lenient towards Samsung re the 840 EVO old data issue, presumably because Samsung built up such a decent rep 'store' (if I can call it that) via their good older products like the 830 and original 840.

By contrast, the hate piled on OCZ long after any of the original-series Vertex complaints were relevant is just nuts, which is a shame because their later products like the Vertex4 and Vector were really good (heck, I have dozens of Vertex2Es and Vertex3s, all running fine). OCZ did make it all worse by a poor support response at the time, but that's not relevant anymore either. People need to move on.

I fully get that people are inclined to not touch a vendor again after a bad experience, but there's taking it to extremes. As you say, if one held this flag for all vendors that mess up, we'd end up not being able to buy anything from anyone.

My speciality is SGIs; I've seen numerous terrible Seagate, Fujitsu, Hitachi and IBM models, but likewise all these vendors have made good drives aswell.

Mind you, talking to a guy at a data recovery company way back, it seems that a lot of vendors get up to all sorts of tricks with disk manufacturing. The guy said one would hardly believe the things they find inside disks which fail sometimes, all sorts of dodgy hand-done fixes to issues so that drive can pass tests & be sold. A common trick is to disable one or more bad platters and sell the drive as a lesser capacity model, bit like CPUs with fewer cores via having some bad cores disabled (by disable I mean the platters still spin along with the others, but the controller board just can't see them, and the relevant heads are not present).

Ian.

I agree.
A company should not have a opinion based on one or two products, it seems like all of them have had something go wrong at some stage or other.
Like I purchased a pile of SSD's off OWC and had quite a few of them fail. OWC were easily contacted and payed for the postage both for the RMA returns and a complete replacement set. Explanation was a bad batch. Fast, easy and I was not out of pocket. I would buy off them again as I have confidence in them.
So I base it more on the customer support than on the products themselves.
I have probably said it earlier, but I have a saying -
As we are but mere humans, mistakes are inevitable. It should not be the mistake that defines us, but how we deal with it.
As such, there are a lot of companies that make it hard to contact, take their time in answering queries and problems (some not at all), and try to avoid their responsibility. In Australia, we have a very well defined set of laws concerning consumer rights, but even then, many try all sorts of techniques to bypass them. One good example is companies say they will not accept a item sent in for warranty unless it is in it's original packaging, but this is contrary to the law. Unfortunately, they con a lot of people when they say it. Asus tried this with me when I had some problems with their motherboards and it took me over a year to finally get a result (even if it was a bit late by then). So until they change their ways, I will not buy their products, it does not matter how good the product is, their reputation has been ruined by their woeful customer service.