SanDisk Announces 4TB SSD

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
SanDisk's new 4TB SSD sure looks mighty impressive.

The Optimus MAX 4TB SSD is the industry’s first SAS SSD to deliver high-density data storage and efficient data throughput for enterprise, cloud and virtualized data centers, surpassing all maximum capacity SAS SSDs and SAS HDDs on the market today.
Click to expand...
 
With a bit of Googling, I found the expected price will be $2500.

By comparison (though these really don't compare) the Samsung EVO 1TB is $460 (x4 = $1840), and your typical 4TB SAS HDD is around $275.
 
It's expensive but I would not mind having a few of these.
 
evilsofa said:
With a bit of Googling, I found the expected price will be $2500.

By comparison (though these really don't compare) the Samsung EVO 1TB is $460 (x4 = $1840), and your typical 4TB SAS HDD is around $275.
Click to expand...

And if one drive craps out, you still have 3 you can use. I'd prefer this option over one 4TB drive.
 
This would probably ease my mind endurance wise. Then again anything above 1TB in the consumer world will probably be TLC.
 
Liger88 said:
This would probably ease my mind endurance wise. Then again anything above 1TB in the consumer world will probably be TLC.
Click to expand...

I am not so worried about endurance of even consumer drives this big even with TLC when you see expected lifetimes in decades assuming 10+ GB of writes daily.
 
My personal choice, would be SSDs with less push for throughput, keeping the same access times and pushing more for capacity. Having tried a few out on my systems from lower end SSDs to the highend, there was such a small difference that if I did not have them side by side and had to uninstall and install the new drive, I never would have been able to tell outside of a stopwatch. I would be fine with M500 speeds and moving on to bigger/cheaper for the time being, we are already getting to the price points to where my next build, even for storage drives I will probably be 100% SSD.
 
If I could afford about 20 on these, I could consolidate all my data/apps/VM's into 2 servers.

If I was building out a new office, it might make sense due to the cost saving of a much smaller computer room, that would need a lot less air conditioning and UPS capacity.
 
I was wondering when these would be coming out....

1 TB and above....


I have two 256 GB SSD drives and they are not full yet.
 
I'm just glad we're starting to see more high capacity SSDs. Even if this one is insanely expensive, it's a good sign for the future.
Those 1TB SSD's are starting to be more reasonably priced. If I create a new build in the next 2 years I'm definitely going that route.
 
Early adopter fee in full effect. Give it about 2-4 years and we'll be seeing sub-$500 pricing on these. Same thing that happened with spindle drives so long ago. Rinse, repeat.
 
evilsofa said:
With a bit of Googling, I found the expected price will be $2500.

By comparison (though these really don't compare) the Samsung EVO 1TB is $460 (x4 = $1840), and your typical 4TB SAS HDD is around $275.
Click to expand...


Hey - where can I get 4tb SAS drives for $275????
 
What I find interesting, is that SSD have easily surpassed mechanical drives in storage density.
4TB SSD = 2.5”
4TB mechanical drive = 3.5”

I recently installed a 1TB mSata ssd in laptop. Less than ¼ the size of a standard 2.5” drive.
Yet, the best I can do with a mechanical 2.5” laptop drive is 1TB for 7200RPM, or 1.5TB for 5400RPM.

I expect we'll see low cost consumer SSD's in the 2TB range this year.
 
MarkL said:
Hey - where can I get 4tb SAS drives for $275????
Click to expand...

I just searched on amazon for "western digital 4tb sas" and got this as the first hit.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
nutzo said:
What I find interesting, is that SSD have easily surpassed mechanical drives in storage density.
4TB SSD = 2.5”
4TB mechanical drive = 3.5”

I recently installed a 1TB mSata ssd in laptop. Less than ¼ the size of a standard 2.5” drive.
Yet, the best I can do with a mechanical 2.5” laptop drive is 1TB for 7200RPM, or 1.5TB for 5400RPM.

I expect we'll see low cost consumer SSD's in the 2TB range this year.
Click to expand...

Yep. We've hit a major form factor barrier with mechanical disks. As we need to go more dense it will be with flash and the successors to flash. It won't be long before you'll be able to buy 100TB of capacity with 1M IOPS in 2U of rack space. No way to do that with mechanical disks.
 
NetJunkie said:
Yep. We've hit a major form factor barrier with mechanical disks. As we need to go more dense it will be with flash and the successors to flash. It won't be long before you'll be able to buy 100TB of capacity with 1M IOPS in 2U of rack space. No way to do that with mechanical disks.
Click to expand...

I'd venture to guess that we could be able to get more capacity than that in only 1U of rack space utilizing large capacity mSATA or uSATA drives...now if only there was a company making 1U RAID racks that could house them. :p
 
DejaWiz said:
I'd venture to guess that we could be able to get more capacity than that in only 1U of rack space utilizing large capacity mSATA or uSATA drives...now if only there was a company making 1U RAID racks that could house them. :p
Click to expand...

Return of the BigFoot drives. We could have about 12tb and 300MBs on the outside edge using current technology. Still sour about that seek time...
 
You could build an 88TB RAID 6 box in 2U for around $70k. Call it 80TB with two hot spares. Works out to 1.6 Petabytes in a rack with room for power?

24 x 4TB SSDs
16 cores
384GB memory
Quad 10 gig ethernet ports

Edit:
$85k for 24 cores, 768GB of memory. That's allot of VM potential in 2U.
 
Last edited:
SAS drives are rarely used alone, these will be bought in quantity when bought.

BlueFireIce said:
My personal choice, would be SSDs with less push for throughput, keeping the same access times and pushing more for capacity. Having tried a few out on my systems from lower end SSDs to the highend, there was such a small difference that if I did not have them side by side and had to uninstall and install the new drive, I never would have been able to tell outside of a stopwatch. I would be fine with M500 speeds and moving on to bigger/cheaper for the time being, we are already getting to the price points to where my next build, even for storage drives I will probably be 100% SSD.
Click to expand...

Manufacturers don't spend that much money on speed, and what they spend is mainly dedicated to compensate for the fact that each NAND shrink deteriorates performance (and durability).

DejaWiz said:
Early adopter fee in full effect. Give it about 2-4 years and we'll be seeing sub-$500 pricing on these. Same thing that happened with spindle drives so long ago. Rinse, repeat.
Click to expand...

Not that easily (2 years, 4 might be possible). Hard drives are mainly technology (R&D) driven, the components are not very expensive, especially the disks are just glass or aluminum plates coated in iron oxide. On the other hand SSDs are made of transistors on silicon.

There's also a limit on where NAND can go and it's being approached fast.

nutzo said:
What I find interesting, is that SSD have easily surpassed mechanical drives in storage density.
4TB SSD = 2.5”
4TB mechanical drive = 3.5”

I recently installed a 1TB mSata ssd in laptop. Less than ¼ the size of a standard 2.5” drive.
Yet, the best I can do with a mechanical 2.5” laptop drive is 1TB for 7200RPM, or 1.5TB for 5400RPM.

I expect we'll see low cost consumer SSD's in the 2TB range this year.
Click to expand...

Size and cost have not much to do with each other. In the case of hard drives it costs more to reduce the size of disks because higher densities are more difficult to obtain. In the case of SSDs shrinking transistors is very difficult and expensive to achieve, but on the other hand when done it means you get more transistors on less silicon, so it ends up cheaper.

I would also bet that hard drive manufacturers have voluntarily slowed down progress in the last few years because they can milk us at current prices and capacities, SSDs have already won performance wise, but can't yet compete in the storage of large data that doesn't need the performance.
 
Aesma said:
Manufacturers don't spend that much money on speed, and what they spend is mainly dedicated to compensate for the fact that each NAND shrink deteriorates performance (and durability).
Click to expand...

Didn't really think they did, but my personal choice, even if performance was static for a little while, would be for size/price right now. We will get there and prices in tech terms have been dropping like a rock since SSDs became the thing. I did not think I would be getting 250-500GB SSDs for the prices they are now this soon. Just ordered a second for my desktop and another for my second laptop, I have not even looked at mechanical drives in some time.
 
Aesma is about the only sane person after that spread of bullshit about pricing.

Flash is made in fabs that are incredibly expensive to build and getting very close to their limit. Samsung is opening a 10nm fab this year, and they will probably open 7nm about 2-3 years after that. and 4nm about 2-3 years after that. That's when we hit the limit of the atom that isn't likely to be surpassed for a few hundred years if ever.

Every process shrink comes with about 30% more transistors in the same space, and 30% less endurance for Flash. So by 4nm we will only be at most seeing 2x more dense chips then current TLC. The price per chip (wafer area) stays relatively stable. So SSD's in 10 years should reach 25 cents per GB at the lowest. So 4TB wont be getting sub $1000 anytime soon.

Weve had SSD's fairly popular for about 6 years now and only seen the price per GB drop from $1.00 to $.50 ...

http://en.wikipedia.org/wiki/5_nanometer < discusses how this is the "last" shrink. although a few working transistors have been made at 4nm, 3nm
 
Last edited:
evilsofa said:
With a bit of Googling, I found the expected price will be $2500.

By comparison (though these really don't compare) the Samsung EVO 1TB is $460 (x4 = $1840), and your typical 4TB SAS HDD is around $275.
Click to expand...

And this is an Enterprise drive as well.. so no comparison really to the EVO....

magixx said:
Seems to be useful only in very specific applications.
Click to expand...

Useful for many things

lets see, you could replace most 6 drive + raid 10 arrays with 2 of these and it would be faster than most 15k spindles......

Spare, power, heat, anywhere you need fast I/O and space....
 
cashkennedy said:
Weve had SSD's fairly popular for about 6 years now and only seen the price per GB drop from $1.00 to $.50 ...
Click to expand...

32GB SSDs back in mid to late 2009 ranged between $200-500. They were $1000+ when they first came to the consumer market in 2007. Before that, they were $3000+ for excruciatingly small capacities and typically only for the Enterprise segment. What was that about $1.00 per GB back then, again?
 
cashkennedy said:
Weve had SSD's fairly popular for about 6 years now and only seen the price per GB drop from $1.00 to $.50
Click to expand...

Uh... no. SSD price per GB only hit $1 in mid-2012:

ssd_price_per_gb-640x343.png


Six years ago, in 2008, we were seeing 32GB SSDs for $500 and 64GB SSDs for $1100. That's $15 and $17 per GB. Most SSD price history charts end at 2010. The best one with older data I found, and it's a good one, is this:

hdd-ssd.0022.jpg
 
^ outstanding! This is going to drive extremely large capacity drives into very affordable price territory.
 
If you look at the second graph of ssd price per gb you see it sitting around 1-2 dollars since 2009 (graph is hard to interpret super precisely because of scale). (EDIT: after looking at the graph again it looks like the arrow saying $1/gb is pointing at a point and not the dotted line, graph doesn't explain what the dotted line is, looks like $2.)

Those graphs are probably average selling price, which in 2009 was a lot more intel drives. If intel was still making up 90% of the market the current price would be over 1gb still in 2014. Plus in 2009 a lot of people were buying the least economical size (32gb) to get a taste of a ssd.

Anyways I was only moderately exaggerating to stir up shit. But that's a pretty damn flat graph since 2009. I truly think your looking at 6-10 years to get to 25 cents a gb ($1000 for a 4tb drive). Based on 2-3 years between node shrinks and 30% bit density gains at each refresh. which are most companies optimistic best estimates.
 
The hope is either a new technology replacing NAND flash (there are candidates, but how many years to get them up to snuff ?), or a way to make production cheaper and cheaper at the same shrink, combined with stacking more and more chips on top of each other, since size isn't really a constraint.

As for hard drives we just saw with the 6TB 7 platters Helium drives that there are possibilities if you want to do it, the problem seems to be that HDD manufacturers don't want to compete at the high end it seems.
 
evilsofa said:
More info about SanDisk's future plans here:

www.computerworld.com/s/article/9248070/SanDisk_announces_4TB_SSD_hopes_for_8TB_next_year

They expect to release an 8TB SAS SSD next year, and to double capacity in the 2.5" SSD size every 1 to 2 years after that. So, 8TB in 2015, 16TB in 2016 or 2017. That goes way beyond anything that the HDD market is planning on doing in data density.
Click to expand...


They're going to blow their 3D Stacking load before NAND's replacement becomes viable. Sounds like a bad business model and even worse if they can't deliver (all smoke and hot air) now. HDD manufacturers are making a killing going slowly by 1TB increments a year.
 
Liger88 said:
They're going to blow their 3D Stacking load before NAND's replacement becomes viable. Sounds like a bad business model and even worse if they can't deliver (all smoke and hot air) now. HDD manufacturers are making a killing going slowly by 1TB increments a year.
Click to expand...

Maybe, maybe not. You don't see these huge SATA drives in most enterprise shops. Too much risk. Too low performance. Just too many concerns there. Flash lets us do cooler things with resiliency thanks to 0ms access time. We can do more than RAID without taking a huge performance hit.

The days of the spinning disk is coming to an end quickly. You get anywhere close to similar pricing and they are dead in the enterprise space. We don't build arrays and size for capacity..we size for performance and usually capacity is already there thanks to larger drive sizes.

We're already starting to see this squeeze. AFAs (All Flash Arrays) are taking off VERY quickly and not just in the high end shops. They are stupid simple to operate and manage...no tuning... no RAID groups. Better resiliency. And the price per GB isn't that terrible given dedupe and compression.
 
This is great. Hard disk vendors have manipulated the market, blaiming on the thailand flooding crisis. In fact, they shipped more disks than ever that year, and broke all revenue records. There never was any disk shortage. If it were not for SSDs, the hard disks would still stuck at 2TB disks for a high price. The reason this oligpoly occurs, is because the fourth vendor was bought and the three left has formed an oligpoly. They have tricked us customers to pay noose bleeding prices for disks. There should be 8-10TB disks now, for a fraction of the price.

Here are several articles about the oligoploy, read this four part series:
http://news.softpedia.com/news/Enjo...-Posts-Huge-Profits-Again-Part-2-283199.shtml
 
NetJunkie said:
Maybe, maybe not. You don't see these huge SATA drives in most enterprise shops. Too much risk. Too low performance. Just too many concerns there. Flash lets us do cooler things with resiliency thanks to 0ms access time. We can do more than RAID without taking a huge performance hit.
Click to expand...

I'm not an enterprise IT guy, but I think I need to comment here. For enterprise, particularly doing e-commerce, response time is an issue, and that point can justify the extra cost of SSD. For consumers, particularly for home media systems, or for manufacturers of "media" products like set-top boxes, spinning disks would probably be preferred for raw capacity and for price/MB.

The days of the spinning disk is coming to an end quickly.
Click to expand...

Do you have any market data to back up this statement?

You get anywhere close to similar pricing and they are dead in the enterprise space. We don't build arrays and size for capacity..we size for performance and usually capacity is already there thanks to larger drive sizes.
Click to expand...

IF ...

Sure, if SSDs could even "almost" match HDDs in price, then HDDs would be as dead a technology as mag tape. But we're not there yet. Sandisk just announced at 1 TB SSD, at "only" $2500 (per another recent posting.) Gee, I just bought (for home use) at 7200 rpm HItachi 4 TB drive for about $170. As an enterprise software sales guy once told me, "Hope is not a successful business strategy."


We're already starting to see this squeeze. AFAs (All Flash Arrays) are taking off VERY quickly and not just in the high end shops. They are stupid simple to operate and manage...no tuning... no RAID groups. Better resiliency. And the price per GB isn't that terrible given dedupe and compression.
Click to expand...

Again, where are the market numbers? AFAs are a niche business. If you start small, you get very impressive growth rates for a few years. But at "maturity," how much will AFAs displace HDD storage in enterprises? I don't know the answer. Do you have the market forecasts, or are you just speculating here?

Dedupe and compression don't depend on SSDs being used, AFAIK.

Are you saying that SSDs are so much more reliable than HDDs that redundancy is not needed? Which IT director is going to bet his job on that assumption, just to save his company a few shekels?
 
x509 said:
Sure, if SSDs could even "almost" match HDDs in price, then HDDs would be as dead a technology as mag tape.
Click to expand...

Small nitpick: Tape is only dead in the consumer market. It's still alive and well everywhere else for large volume backups and data archiving.
 
techrat said:
Small nitpick: Tape is only dead in the consumer market. It's still alive and well everywhere else for large volume backups and data archiving.
Click to expand...

Thanks. I stand (properly) corrected on this point. It's my * impression * that even in the enterprise, tape is less important than it used to be.
 
x509 said:
Thanks. I stand (properly) corrected on this point. It's my * impression * that even in the enterprise, tape is less important than it used to be.
Click to expand...

There was a big buzz in the office lately when Sony announced that they developed a crystal magnetic tape that can store 185TB.

Large volume still wins when it comes to tape vs HDs.
 
Disclaimer: I work for a VAR that sells a lot of EMC storage gear. In fact, I'm on a plane back from EMC World right now. Spent all week in product and roadmap briefings.

x509 said:
I'm not an enterprise IT guy, but I think I need to comment here. For enterprise, particularly doing e-commerce, response time is an issue, and that point can justify the extra cost of SSD. For consumers, particularly for home media systems, or for manufacturers of "media" products like set-top boxes, spinning disks would probably be preferred for raw capacity and for price/MB.
Click to expand...

Very few companies are in the media business. Most just have applications that help run the actual business itself. Logistics, accounting, customer management, internal IT like email, etc. They all benefit from fast I/O and are often not overly capacity driven.

Do you have any market data to back up this statement?
Click to expand...

Go read any Gartner or IDC report you want. I'm not saying they'll be gone in 18 months but in 3 years that majority of storage sold will be on flash. There will be archival cases where 10TB spinning disks make sense...but those will become more and more the exception as flash gets bigger and cheaper. That trend is absolutely starting right now. Almost every customer I deal with that is currently refreshing storage is weighing the option of going all flash..or at least very heavy flash. That won't reverse.

IF ...
Click to expand...

I don't need a 4TB SSD to be the same price as a 4TB spinning for this transition. Do you have any idea how many 600GB drives are sold in enterprise arrays today? A LOT. Because it takes a lot of them to get performance needed and 600GB x "a lot of disks for performance" gives them the capacity they need. That's why with storage using spinning disks we rarely size for capacity...once we size for performance we're often there. That changes with flash. Instead of 100 x 600GB to get what you need I can do 25 x 800GB SSD and it's enough..and the performance is 30x the spinning config.


Sure, if SSDs could even "almost" match HDDs in price, then HDDs would be as dead a technology as mag tape. But we're not there yet. Sandisk just announced at 1 TB SSD, at "only" $2500 (per another recent posting.) Gee, I just bought (for home use) at 7200 rpm HItachi 4 TB drive for about $170. As an enterprise software sales guy once told me, "Hope is not a successful business strategy."
Click to expand...

Go price a 1TB drive in a real storage system. They aren't $170. A 1TB SSD isn't $2500 either..but it's not as skewed.

Again, where are the market numbers? AFAs are a niche business. If you start small, you get very impressive growth rates for a few years. But at "maturity," how much will AFAs displace HDD storage in enterprises? I don't know the answer. Do you have the market forecasts, or are you just speculating here?
Click to expand...

AFAs were niche 5 years ago when it was Violin and RAMSAN with very purpose built products. We're no longer there. I'm not speculating I sell it and again, go look at Gartner and IDC. HD capacity growth has been outrunning data growth in companies for a while. You just don't see these big drives sold unless it's a long term, VERY low performance archival use case. But a 20TB EMC XtremIO brick that can do anywhere from nothing to 10:1 dedupe giving you 200K real world no BS IOPS solves a lot of problems for reasonable money.

Dedupe and compression don't depend on SSDs being used, AFAIK.
Click to expand...

Never said it was. What I said was that by using dedupe and compression I can make an AFA even more reasonable. Sure..you can do it with spinning disks too but..and again...we rarely size for capacity. It's usually not needed.

Are you saying that SSDs are so much more reliable than HDDs that redundancy is not needed? Which IT director is going to bet his job on that assumption, just to save his company a few shekels?
Click to expand...

I never said redundancy wasn't needed. I said things like RAID, stripe size tuning, and manual disk group configs were dead. RAID has outlived its usefulness. There are better ways to protect data. Better systems. Things that let me do rebuilds much faster...let me fail more drives. Example..and I use this because I deal in EMC... The XtremIO AFA doesn't do RAID. It uses a new system. With it you can fail up to 5 drives in a shelf (5 out of 25). Can do a drive rebuild in 14 mins (lab tested by me on a 10TB brick). Can maintain that 200K IOPS performance even in a failure situation. RAID is on the way out.
 
You must log in or register to reply here.
Back
Top