No announcements/reviews from/for WD or Seagate lately?

[mikeymikec]

Have I missed or forgotten something, or do other people agree that WD and Seagate have both been very quiet, possibly since the flooding?

I can understand that R&D would take some sort of hit because of the flooding, but I would expect both manufacturers would get right back on R&D as soon as they were able.

It seems to me that in past years one might have expected a new addition to the Barracuda series (from 7200.7 to .8 I mean) about every year or 1.5 years, and the differences in each new version have been quite modest (as has HDD development been for as long as I've been interested).

Of course HDDs are going to get supplanted by SSDs in the mainstream arena, but I doubt that HDDs are going to disappear completely just like tape drives haven't (my point is that there tends to be an ideal application for a technology for quite a while past its "current tech" era), and I strongly doubt that WD and Seagate are just going to give up and die because of HDDs being supplanted.

Seagate have done a bit of hybrid work, but I would have thought there would have been more progress along that line and/or some flat-out buying of an SSD maker / chip maker to throw their lot in to SSD development.

I can't remember the last time I read a review for a new hard disk or anything from WD/Seagate. I find this a bit odd. A quick trawl through Techreport's and Anandtech's storage reviews came up with 3 reviews on each site in two years, (excluding NAS kit) and one was for a Hitachi disk.

 

[Elixer]

What are you looking for, they are just starting 4TB HDs now...

 

[mikeymikec]

Progress, new products etc? Capacity isn't the only obstacle to overcome. My feeling is that new products (and reviews for them) seem to have dried right up from either manufacturer, in comparison to say 2009 or earlier.

Example: I googled announcement dates for each new number in the Barracuda 7200 series:

7200.7 announced 2002
7200.8 announced 2004
7200.9 announced 2005
7200.10 announced 2007
7200.11 announced 2007
7200.12 announced 2009
7200.13 (Referred to as Barracuda XT) announced 2009
7200.14 announced 2012

Considering that 12 and 13 were both enormous capacity drives, none of the "mainstream" sizes of say 1TB or below, it's like time has stopped for those instead of the incremental improvements that used to happen between series. Every series up to and including 12 had a pretty comprehensive range of capacities available, then that stopped.

I think the flooding that affected both manufacturers occurred in late 2011, so that doesn't account for the gap.

I can't think of a good way of analysing WD's product releases since they adopted the "Caviar Blue/Black/Green" style wording.

 

[blastingcap]

HDD makers know that SSDs will come to dominate smaller capacities, and that their stronghold will remain high-capacity drives for home/SOHO consumers and commercial servers. I don't think SSD is cost-competitive in a server environment yet, and probably will not be for a long time to come, because of NAND wear issues exacerbated by node shrinks. There are some technologies to extend NAND life, so you can use cheaper or smaller-node NAND and retain the same life expectancy, but NAND write life expectancy still pales compared to spinners.

 

[mikeymikec]

@ blastingcap

I agree with what you're saying, though I think it'll be at least another two years before OEMs start typically shipping computers to the mainstream with SSDs in, and two years' worth of OEM storage sales is not something to be taken lightly. The more marketing noise that SSD manufacturers make will trim down the amount of time that HDDs have left in the mainstream (assuming that they don't start doing some serious under-cutting to grab that space sooner).

 

[Hellhammer]

I think the biggest reason why we haven't seen any major releases lately is because there isn't much market for drives bigger than 2TB, at least from what I have noticed. There is not much performance benefit from going bigger and you risk losing TBs of data at once if the drive fails. People who need TBs will rather buy multiple drives and run them in RAID 5 for instance to get redundancy.

HDD makers know that SSDs will come to dominate smaller capacities, and that their stronghold will remain high-capacity drives for home/SOHO consumers and commercial servers. I don't think SSD is cost-competitive in a server environment yet, and probably will not be for a long time to come, because of NAND wear issues exacerbated by node shrinks. There are some technologies to extend NAND life, so you can use cheaper or smaller-node NAND and retain the same life expectancy, but NAND write life expectancy still pales compared to spinners.

SSDs are already widely used in the enterprise segment. Enterprise SSDs typically use either eMLC or SLC NAND to overcome the endurance issue. SSDs have the advantage that their failure can be predicted using SMART data, HDs just die without a notice.

You also don't have to go SSD-only, many SSD OEMs talk about tiered systems where the most frequently data is stored in SSDs while the other data resides in HDs. That's similar to SSD caching, although there may be additional tiers (for example PCIe SSDs with SLC NAND for the tier 1 data, then SATA/SAS SSDs with eMLC for tier 2 and finally HDs for tier 3).

 

[blastingcap]

I dunno, I know of people who are bay-limited and would like greater data density. In a recent NAS build, I had four bays, and wanted to go with 3TB drives to fill them rather than 2TB drives, for that reason. But I also have over 28 TBs of hard drives and over 1TB of SSDs, so I'm probably not representative of most people.

I thought that the market for enterprise SSDs was niche, but the "cache" model makes sense and could make it less-niche. So I guess SSDs and HDDs will coexist in that market segment.

I can definitely see the day when you have cheap TLC NAND with some fancy tricks to extend wear life, a few shrinks from now, dominating the small-capacity market though. Was not saying that's going to happen overnight or even in the next couple of years. I'm guessing 2017 but could be way off.

I think the biggest reason why we haven't seen any major releases lately is because there isn't much market for drives bigger than 2TB, at least from what I have noticed. There is not much performance benefit from going bigger and you risk losing TBs of data at once if the drive fails. People who need TBs will rather buy multiple drives and run them in RAID 5 for instance to get redundancy.



SSDs are already widely used in the enterprise segment. Enterprise SSDs typically use either eMLC or SLC NAND to overcome the endurance issue. SSDs have the advantage that their failure can be predicted using SMART data, HDs just die without a notice.

You also don't have to go SSD-only, many SSD OEMs talk about tiered systems where the most frequently data is stored in SSDs while the other data resides in HDs. That's similar to SSD caching, although there may be additional tiers (for example PCIe SSDs with SLC NAND for the tier 1 data, then SATA/SAS SSDs with eMLC for tier 2 and finally HDs for tier 3).

 

[Elixer]

SSDs are already widely used in the enterprise segment. Enterprise SSDs typically use either eMLC or SLC NAND to overcome the endurance issue. SSDs have the advantage that their failure can be predicted using SMART data, HDs just die without a notice.

Is that only true for enterprise level SSDs ?
I have had 2 SSDs die with 0 indications from SMART that it was about to fail.
I have seen quite a few HDs have SMART indicators show up before they go belly up.

 

[blastingcap]

Is that only true for enterprise level SSDs ?
I have had 2 SSDs die with 0 indications from SMART that it was about to fail.
I have seen quite a few HDs have SMART indicators show up before they go belly up.

I think he may be referring to some companies installing wear indicators, which theoretically tell you when NAND has reached the end of its expected life, but in reality drives can die prematurely or survive long after their rated lifespan, as the testers at XS have found: http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm

Anyway, does it matter? Any decent enterprise system will have redundancies such as RAID6 or mirroring, regardless of whether they are using solid state or spinners or both. So I wouldn't include wear indicators as a big advantage for SSDs. But maybe hell meant something else re: SMART.

 

[Vinwiesel]

If I had to guess I would say they are transitioning from traditional platters to Heat-Assisted methods, so there may be a lag in releases until the new tech goes mainstream.

Was there a lag back when they switched to perpendicular recording?

 

[mikeymikec]

Perpendicular came in with 7200.10 I think.

I just searched a mailing list I'm on and it is first mentioned in June 2006.

 

[Railgun]

What`s there to analyze? Something you need that they don`t do? Hoping for something in particular? 4TB not large enough?

You`re creating speculation where none needs to be made.

The short of it is, as a consumer...so what?

 

[mikeymikec]

What`s there to analyze? Something you need that they don`t do? Hoping for something in particular? 4TB not large enough?

You`re creating speculation where none needs to be made.

The short of it is, as a consumer...so what?

If you think that the only thing to have changed over the years is capacity, then perhaps you ought to try digging out a UDMA 33/66 IDE 40GB disk and run Win7 on it.

If you weren't interested in this topic, why read it or respond?

@ Vinwiesel

Thanks for that, I found this in the process:

http://www.seagate.com/about/newsroom/press-releases/terabit-milestone-storage-seagate-master-pr/

 

[BFG10K]

SSDs have the advantage that their failure can be predicted using SMART data, HDs just die without a notice.

Actually it's the opposite. In the vast amount of cases when SSDs die, they die suddenly and without warning, and are totally bricked. SMART can't predict that.

With HDDs you’ll often get warning signs (e.g. large amounts of reallocated sectors, slow performance, and even abnormal noise), giving you enough time to pull the data.

 

[fuzzymath10]

If you think that the only thing to have changed over the years is capacity, then perhaps you ought to try digging out a UDMA 33/66 IDE 40GB disk and run Win7 on it.

If you weren't interested in this topic, why read it or respond?

@ Vinwiesel

Thanks for that, I found this in the process:

http://www.seagate.com/about/newsroom/press-releases/terabit-milestone-storage-seagate-master-pr/

Funny enough I think the difference is not as great as the capacity difference suggests. In most cases other than sequential, you're in single MB per second. I've installed vista and win7 on 40gb 5400rpm laptop drives and it was just slightly more painful than a contemporary 1TB+ 7200rpm disk because you're stuck with random ops most of the time. You need an SSD or hybrid drive before you notice a difference. Once superfetch finally finishes reading into memory then drive speed matters less but that process can take a long time.

 

[mikeymikec]

Actually it's the opposite. In the vast amount of cases when SSDs die, they die suddenly and without warning, and are totally bricked. SMART can't predict that.

With HDDs you’ll often get warning signs (e.g. large amounts of reallocated sectors, slow performance, and even abnormal noise), giving you enough time to pull the data.

However, in time I think when SSDs have been around in the mainstream for a while that their quirks will be learnt in the same way that the quirks of HDDs have been learnt. HDDs can die suddenly and completely unexpectedly too IMO. A new implementation of something SMART-like for SSDs will probably be developed too (if it hasn't already, I can't say I'm very knowledgeable about SSDs), I'm sure of that.

Funny enough I think the difference is not as great as the capacity difference suggests. In most cases other than sequential, you're in single MB per second. I've installed vista and win7 on 40gb 5400rpm laptop drives and it was just slightly more painful than a contemporary 1TB+ 7200rpm disk because you're stuck with random ops most of the time. You need an SSD or hybrid drive before you notice a difference. Once superfetch finally finishes reading into memory then drive speed matters less but that process can take a long time.

True to a point, but for example someone I knew had a very similar computer build to mine (same board, same processor, enough RAM in both, both XP, same AV), except that they hadn't replaced their IDE 40GB 7200.7 disk and I had replaced my disk with a SATA one somewhere along the way (the board's SATA chipset didn't have NCQ IIRC). OpenOffice (same version) opened on mine about 10 seconds quicker than it did on his. And it wasn't due to anything like a disk issue or fragmentation or a dodgy install/OpenOffice profile information / add-ons, etc.

Also, how my disk handles multiple loads is a heck of a lot better from its predecessor, a 400GB SATA one on the system I already mentioned. I still sometimes see transfer speeds of say 30MB/sec even while something else that is regularly hitting the disk is going on. On the previous setup (admittedly an entire platform change between my current 500GB and that system with the 400GB disk), I bet throughput would have dropped to about 5MB/sec if I had done something similar at the same time as the transfer.

There have been lots of modest improvements on the way. I've never seen a UDMA100 disk do 100MB/sec (or anywhere near that, perhaps not even 50MB/sec, but it has been a while). I've seen a SATA II disk do 100MB/sec, sometimes a little faster. Packing more data onto each platter will probably make the biggest difference, then there's the cache, the efficiency of the interface, NCQ helps in certain situations IMO, then who-knows-how-many under the bonnet changes that are made with a new model of disk, better controllers, etc.

 

[Hellhammer]

Actually it's the opposite. In the vast amount of cases when SSDs die, they die suddenly and without warning, and are totally bricked. SMART can't predict that.

With HDDs you’ll often get warning signs (e.g. large amounts of reallocated sectors, slow performance, and even abnormal noise), giving you enough time to pull the data.

That might be true for consumer-grade drives but reliability is taken way more seriously with enterprise drives. The premium in price isn't just from more expensive NAND, it's from extensive validation and testing that is required for enterprise. On top of in-house testing, interested customers usually do their own validation or pay for a third-party to do that before they pull the trigger and spend millions on SSDs. That's why product announcements are often made months, even a year in advance; samples are available for validation but mass availability is still far away.

So yeah, while consumer SSDs can fail unpredictably, it's far more unlikely that the same will happen to an enterprise SSD. Sure there is always a fraction that fail but that percentage is extremely small.

The problem with HDDs is that you can't predict it's endurance. Take two drives and one may fail after 1PB of writes, whereas the other may be good for 10PBs. With SSDs you have a pretty good idea of when the drive will wear out and when should it be replaced.

The real reason why SSDs are used in enterprise is the performance, though. One PCIe SSD can easily outperform 20 HDs, which can reduce the total amount of nodes, hence lowering power consumption and requiring less space. Of course, it all depends on what is the usage; SSDs won't help if the system is limited by processing power.

 

[Railgun]

If you think that the only thing to have changed over the years is capacity, then perhaps you ought to try digging out a UDMA 33/66 IDE 40GB disk and run Win7 on it.

If you weren't interested in this topic, why read it or respond?

What's your point? It will still run. Yes it will be slow but it will still run. Are you trying to suggest there should be some quantum leap in HDD tech to compete with SSD? All you're asking is why is there no news. Who cares? What are you expecting?

My interest is always around the progress of technology, but the fact that some manufacturer hasn't met someone else's expectations with no explanation to what those expectations are seems somewhat frivolous.

 

[mikeymikec]

Ok Railgun, either read what I wrote and respond rather than try to read something that blatantly isn't there. Your first two questions I've already answered. Your third question you've already asked and I've answered, and you shouldn't have needed to ask the fourth question as the answer to it was already implied in the original post.

If SSD manufacturers went silent for 2-3 years from today and little changed (including cost) except to produce say a 1TB SSD for a large amount of money, I think most people would be scratching their heads and wondering what was going on. Anandtech and other such sites probably have a 20:1 ratio of SSD to HDD articles currently, I personally would think it was a bit odd if that ratio dropped to 1:1. Whether you would be still standing around saying "what did you expect? Your 32GB SSD still works doesn't it? Who cares?" are questions I'm not much interested in as you evidently have nothing to say.

 

[Railgun]

I'm reading it....I don't see a question. I don't see a concern. I have no idea what you're getting at. All I see is a thread being created for the sake of creating a thread.

You said..

I can't think of a good way of analysing WD's product releases since they adopted the "Caviar Blue/Black/Green" style wording.

What are you trying to analyze? If you're a shareholder that actually has a worthwhile stake, ok. As a consumer...again...what are you trying to get? Apple did a refresh much sooner than they normally do with the iPad...what's that mean? Nothing...other than they did it. Same to this...so they may take a little longer to have some incremental release to old tech...ok...so what?

Your thread is trying to come up with something that blatantly isn't there. I'm trying to figure out what you're trying to get out of this thread. There's no profound insight gained by suggesting that some company is taking a little longer to create some new product that has neared the limit of its technology.

 

[mikeymikec]

I wasn't making a conclusion, I was saying how I saw the situation and wondered if anyone knew any more than I did and/or whether they've heard anything interesting going on, whether people agree or disagree that this is an odd trend, etc.

Like (probably) a lot of people here, I have a career in IT and hardware/software advancements are of personal and professional interest.