Google Proposes New Hard Drive Format For Data Centers (thestack.com)

An anonymous reader writes: In a new research paper the VP of Infrastructure at Google argues for hard drive manufacturers and data center provisioners to consider revisions to the current 3.5" form-factor in favour of taller, multi-platter form factors — with the possibility of combining the new format with HDDs of smaller circumference which hold less data but have better seek times. Eric Brewer, also a professor at UC Berkeley, writes "The current 3.5" HDD geometry was adopted for historic reasons – its size inherited from the PC floppy disk. An alternative form factor should yield a better TCO overall. Changing the form factor is a long term process that requires a broad discussion, but we believe it should be considered."

Comments

Form Factor not "Format"

[michaelmalak]

Also, I thought the world was going SSD anyway, which is thinner, not thicker?

Re:Form Factor not "Format"

[drinkypoo]

The world will probably keep using spinning rust until purchase price (not TCO) on SSDs is lower. I wouldn't be surprised if makers went back to 5.25 x half height, and low spindle speeds. It would still permit large throughput with high density, but seeks would be slower. Not a big deal with enough caching in front of them, and/or with enough disks in an array. As SSDs approach HDD price, they will take up more of the workloads that actually have to be fast anyway.

Re:Form Factor not "Format"

[p4ul13]

SSD is the heir apparent, but platter based disk storage will likely provide higher capacity at denser, more affordable prices for quite some time to come. I suspect Google is proposing this altered platter HD design as something that could bridge the gap until SSD reaches an affordability / density point that can catch up / replace conventional platter HD designs.

Re:Form Factor not "Format"

[Anonymous Coward]

Um, it hasn't been "rust" in decades. That's as condescending as saying SSD's are lumps of sand...

Re:Form Factor not "Format"

[Anonymous Coward]

Nope. SSD's are dead.

A Mechanical drive will last 7 years, a SSD used in the same machine will last 2 years, 3 tops. You pick SSD's for low-write high-bandwidth applications, eg streaming video and images. Mechanical drives are for low-bandwidth high-write applications like logging. So things like Google where Youtube is going to file a lot of content into cold-storage, SSD's are a waste, and in fact it may be more practical to just burn discs for content that nobody is watching.

In the case of what Google is proposing, what makes practical sense is to instead of having 5.25" and 3.25" platters, mount 4 2.5" drives into the same space that a single 5.25" drive space takes, perhaps make a new transverse 5.25" mount, so you you fit a 2.5" height with the platters to the depth of the 5.25" bay. So you'd get something that resembles "blade" installations but one PCB instead of 4 separate PCB's.

However the one omission in this proposal of making the drives taller to have more platters in them is that yes theoretically you get twice the capacity if you double the platters (eg 2 to 4) and twice the data bandwidth IOPS, but you also are still pushing this over the same SAS/SATA connection, which means you approach a SSD's lower-end performance.

As for SSD's, they are terrible in a data center environment. Try copying a large file from one SSD to a mechanical drive or to another SSD of weaker performance, the system will, ALWAYS lockup, regardless of the drive because the OS's disk cache will exhaust the ram on the system in doing so. Or at least this is what always happens on Linux. I don't have any SSD's on BSD machines for comparison, but I have a dozen Linux systems running SSD's, and they are the worst choice for servers. You are better off maxing out the RAM in the system and then running various memory caches (eg memcache, varnish) with a mechanical drive than you are trying to use a SSD, because all the performance is at the memory cache. Instead of trying to save money with conservative amounts of RAM, and expensive SSD's, buy cheaper SATA mechanical drives and push 1/4 of the RAM on the system into software caches (not the disk cache itself.)

Don't get me started on how awful cloud services are that push SSD's. The "virtual machine" server is the worst thing to ever take root in a data center. Now instead of buying 10 fast systems with the right CPU, RAM and Disk configuration, you end up provisioning those 10 servers with 80 VM's, one per CPU core and 1GB of ram, and instead under-size 80 virtual machines with the naive assumption that unused performance/ram will be given to the machine that needs it. The reality is, even at Amazon Web Services, they provision the machine so you have 100% of that division of the machine, and any unused performance is just gone and wasted instead. Over provisioning results in poor performance, and that is what the bean counters do not understand. Once you throw SSD's into the mix, you end up wearing out the drives much faster as machines are spun up and down on the same physical storage. So now instead of having SSD's that last 3 years, you get SSD's that last 90 days.

Re:Form Factor not "Format"

[Anonymous Coward]

Quantum Bigfoot anyone ?

Re:Form Factor not "Format"

[gman003]

That's odd, because my laptop's SSD is four years old and still has plenty of usable life left - and it's from a middle-line vendor, from the early SATA3 days, so it's not even a particularly good SSD. The hard drive in the same laptop (dual-bay) is actually reporting as closer to failure. Maybe that's because it's a laptop, so it suffers more vibration and temperature variation, which is harder on hard drives than solid-state.

And the rest of your bitching seems to be based more on shoddy cloud hosts than SSDs, or on badly-configured servers. "SSDs are too fast, they bring down the entire system by filling up RAM"... wouldn't that be true of hard drives as well, IF they could transfer data that quickly?

Re: Form Factor not "Format"

[0xdeaddead]

First thing that came to mind. And that didn't end well!

Re:Form Factor not "Format"

[Jahmbo]

Why not just add a second, third, fourth .... arm and read/write heads, slap some more logic on the board to handle them?

Re:Form Factor not "Format"

[Anonymous Coward]

Thank you. The phrase 'spinning rust' belongs in the dustbin of history along with such pretentious pap as 'vaxen'.

Re:Form Factor not "Format"

[NeoMorphy]

SSDs have been in data centers for years now. They are more reliable than hdd. That wasn't always true, but it is now. You don't use them for "low write/high-bandwidth", you use them for high iops/random access like database indexes, or the entire database if you can cost justify it. I have never seen a linux server lockup when copying files to/from and we have thousands of linux servers, you might need to tune something if that is happening. We have multi-tiered storage with 7200/10000/flash drives. The downsides to ssd are cost/data density, otherwise hdd would be dead, the noise/power/heat/performance difference is huge.

Virtualization is awesome! Servers have been increasing in performance every year while server applications resource usage grows at a slower rate. It's ridiculous to buy a server for a webserver application and another for network filesystems and printers etc. You also have to manage heat dissipation and worry about power/network/fc cables to a bazillion servers. With virtualization you can downsize your server farm(and the network/power/fc cables) 20X+ and save on hardware/cooling/power/management. If a virtual server needs more resources you can vmotion it to another server with more available resources, you can even have the process automated. With san storage you can carve out servers anytime you want. It used to suck when a server died because of hardware failure and you were down until it was fixed. Now you can bring it up in another frame!

Re:Form Factor not "Format"

[lgw]

Old SSDs died quickly under DB loads - not enough write count in their lifetime. New ones are better, but still won't last as long as HDD. This is only going to improve over time, though, and at the right price, who cares about 2 vs 4 year lifetime?

The HDDs you have to use for high IOPS DB load are dammed expensive in the first place: it's the last domain of big-box storage (think 10x consumer drive cost, 100x with fancy replication software built-in).

Google doesn't use that big-box crap of course, but I'm baffled why they want a faster HDD standard that won't come for years - in a few years, will anyone still care about HDD speed?

Re:Form Factor not "Format"

[Anonymous Coward]

How exactly are HDDs denser?
You can buy a 4TB 2.5" 7mm z-height SSD right now. Or a 16TB 15mm.
Go find a HD approaching that capacity/volume.

Re:Form Factor not "Format"

[swb]

I think you're right, but I think it only counts/matters for organizations operating at the spreadsheet analysis scale where the potential savings are really only realized across many thousands of disks in extremely customized environments.

It also wouldn't surprise me if this was also being floated by Google to induce hard disk makers to leverage their existing manufacturing base to mass produce something that really only a very small number of customers are likely to have in any interest in. Hard disks are probably the one component Google can't just design and have made for themselves, unlike the rest of their rack hardware.

I also wonder if the business math on Google's distributed storage system doesn't make as much sense with SSDs versus hard disks and they worry that their model is going to be upset by HDDs potentially getting more expensive as everyone else transitions to SSDs. It doesn't seem too outrageous to think that in 10 years that a lot of the manufacturing capacity of HDD makers will be economically nonviable as demand shrinks. They won't disappear, but as demand shrinks and manufacturing follows, price might go up.

Re:Form Factor not "Format"

[nanoflower]

I agree with you that this only makes sense for very large customers of hard drives. If Google really thinks this is a good idea they should approach one of the vendors with a long term commitment to buy the drives or a large investment for them to develop the drives. That's really the only way I see a Seagate or WD spending their time/money to develop a product that has no retail purpose and may have no commercial customers. Let the HD companies know it won't be a wasted investment and they'll come up with something that will help Google cut down on space/power usage for their HDs.

Re:Form Factor not "Format"

[LWATCDR]

This is more enterprise than consumer. File systems like ZFS can use SSDs as a cache for spinning platters. On a modern server you may have a system that uses RAM as a traditional disk cache followed by an SSD or array of SSDs as a second cache layer, and then disks as the mass storage.
It can even be pretty smart and using and ageing system to move files in and out of the SSD cache based on when they were used last and you could even tag some files to always be in the SSD cache and others to never be in the SSD cache.
My question is simple. Will the benefits of these tall multi-platter drives be worth the cost? Sticking with the tradition drive sizes gives you economy of scale.

Re:Form Factor not "Format"

[Anonymous Coward]

Old SSDs died quickly under DB loads - not enough write count in their lifetime. New ones are better, but still won't last as long as HDD.

Let's take a decent 15k, even ignoring seek times the rotational latency limits us to about 500 IOPS.
Saturating that with 4k writes 24/7 for a year... about 63.12 TB written.
What's the write rating on a 200GB intel DC P3700? 3737.6TB.
Do you honestly expect that HD to survive for nearly 60 years?

Re:Form Factor not "Format"

[swb]

I could see some of the larger SAN vendors getting behind this, if only as a way to keep customers paying top dollar for SSD and tiering features. They would gain an additional way of charging more for less (super magic form factor high performance high density hard disks that only work in our custom enclosures..).

My guess, though, is that they're probably going to see some of their business erode from SSD-only vendors whose products will provide better performance at less cost because they can eliminate some of the overhead associated with tiering and caching schemes and supply lighter weight controllers.

Re:Form Factor not "Format"

[sunking2]

So we keep using hard drives because they are cheaper and somehow raising their costs due to development, retooling, deployment costs, etc is supposed to be a good thing? Raising the costs of something you are using because its cheaper seems the opposite of what you want.

Re:Form Factor not "Format"

[Dutch Gun]

You could very well be right. Speaking of oddball heights, the first 500 *MB* drive I bought (back when the main network drive was 120MB) cost $1000, and it was actually a 3 1/2" double-height size, meaning the bay next to it had to be clear before I could install it. It wasn't a problem since I was simply installing it in a workstation. This obviously wouldn't work for Google, since I'm certain they use computers with front-mounted hot-swappable 3 1/2" drive bays all neatly packed together - I've seen how nicely these work with my Synology 5-bay NAS. Unless a new form factor becomes standardized, you can't really hack in a solution... at least not on the scale Google is dealing with.

I don't think Google is going to get its way here with a new standardized size, at least at mass adoption scales. Inertia is pretty damn hard to overcome, even if potentially superior solutions exist. I mean, the US is still using imperial measurements, for heaven's sake. The fact that we measure them as 3 1/2" inch drives should tell you something about how hard it is to change standards.

Re: Form Factor not "Format"

[U2xhc2hkb3QgU3Vja3M]

The data became blurry?

Re:Form Factor not "Format"

[Anonymous Coward]

you should get out more.... because what you spend your time on is not making you know more and it shows.

Re:Form Factor not "Format"

[Megol]

I guess you call your computer chips molten sand?

Re:Form Factor not "Format"

[kimvette]

Funny.. I use an ssd for vms and swap and although it's thrashed heavily when I fire up all the VMs at once I've lost no capacity and no errors have been reported. Reliability has been so solid I'm thinking of replacing the spindles in my server with SSDs in a RAID5 or RAID6 array.

There are occasional lemon SSD model runs... but that's true of all hard drive manufacturers as well.

Furthermore empirical evidence arising from analyses from quite a few data centers indicate that SSDs are more durable than hard drives at this point. Continued avoidance of the latest-generation SSDs is based on superstition, not evidence.

Re: Form Factor not "Format"

[drinkypoo]

It didn't end well because they were designed to be ultra-low-cost. The goal in this case would be higher reliability and lower power consumption as compared to drives with higher spindle speeds.

Re:Form Factor not "Format"

[Darinbob]

Part of the issue is that the current form factor is wasteful of space just by itself, regardless of speed or storage. Stack more disks, add a second interface, and you can get more platters into the same amount of space and probably with less power.

Re:Form Factor not "Format"

[Coren22]

So, if it isn't Iron anymore, what are the magnetic domains made from?

Re:Form Factor not "Format"

[Coren22]

You can? Where? Biggest I can find in 2.5" is 2TB, so where are you finding 16TB SSD drives?

Re:Form Factor not "Format"

[JohnStock]

> It would still permit large throughput with high density, but seeks would be slower. If you read the article that is the exact opposite of what they want.

Re:Form Factor not "Format"

[JohnStock]

Many analysts are saying starting this year that is no longer true anymore and that starting next year SSD cost/density per unit of storage will be cheaper than HDDs

Re: Form Factor not "Format"

[Anonymous Coward]

The term we use for folks like the poster above is "not even wrong."
See also: "just enough knowledge to be dangerous."
Bonus points if they write out "Period." or "End of discussion." or "Case closed." to make their argument sound "snappier."
But yeah, I've been using and abusing SSDs for years and they still work fine. As expected. Period. Case closed. :-)

Re:Form Factor not "Format"

[Pentium100]

And what's the price difference?

Intel 3700 series SSD, while good, are really expensive. A 100TB server made from them would be really expensive and there is no point in that, where a big RAID10 array of hard drives with 1TB of SSD caching (using 3700 series) is much cheaper. Especially if the server itself is used to store video files.

Re: Form Factor not "Format"

[Pentium100]

That's because the drives were designed to be low cost consumer devices. There were 5.25" Full height SCSI drives for a while before and after the bigfoot. I have one such drive - a 1.2GB drive made in 1992. Still works.

Re:Form Factor not "Format"

[mt2mb4me]

I am an enterprise storage engineer so I am getting a kick.... Seriously though. SSD's even by the big manufaturers are considered a consumable part. They are not covered under any warranty. SSD is great as said earlier for fast read-write applications. What you do on a laptop is absoulutely nothing compared to what is done to a data array with a transnational database on it, like for an ATM network, or search engine. SSDs will always be fast, but they are not as reliable as a 600GB 15k drive, or even a 2TB SAS drive. Facebook, actually stores deep storage to an optical library that pulls up years old pictures only under request. for mid legnth storage, Spindles still rule the roost. As for form factor, I don't see any reason for it to change. they all use the same technology as a standard pc.

Re:Form Factor not "Format"

[lgw]

For high IOPS big-box load, you don't use 1 TB drives though. You use ~500GB 15K SAS drives, and maybe use half the space, because it's all about spindle count.

Enterprise-quality SSDs are already cheaper than big-box storage, but that doesn't really help - you can't replace a FC/iSCSI-attached storage array with some local drives. There are a variety of competing "off-brand" SSD storage arrays, but EMC has really good salesmen and really good FUD.

SSDs will conquer when big, less-technical companies dare to move away from the Oracle/EMC world to a distributed computing model and commodity servers with SSDs for far, far cheaper. But those business application written for Oracle are very sticky.

Google never went there in the first place. They started with a "lots of disposable commodity servers" model, and each server has a far more reasonable IOPS load, well within what local SSDs can do, and well within the lifetime IOPS limit of those SSDs. It's just about cost, and in a few years is there really going to be that much cost difference?

Re:Form Factor not "Format"

[Pentium100]

It all depends on the usage scenario of the storage. For a database, random access is very important. For a big VOD server it's not as important, especially if the cache is big enough. Even 5400RPM 6TB drives are good enough for that (a 34 drive RAID10 array of them that is) as the access is not done in 4K blocks, but in 128K-1M blocks or even bigger.

Re:Form Factor not "Format"

[OrangeTide]

Rust is iron oxide, which is chemically different than metallic iron. And cobalt is one of many ferromagnetic materials besides iron.

Re:Form Factor not "Format"

[8086]

but you also are still pushing this over the same SAS/SATA connection, which means you approach a SSD's lower-end performance.

For non-cached data, I don't see these new 2.5" drives exceeding SATA's 6Gbps (about 600 MBps) or SAS's 12Gbps (1.2 GBps). By the time these get put into production, SATA and SAS 4 (or, who knows, 5) will already be out.

Re:Form Factor not "Format"

[drinkypoo]

> It would still permit large throughput with high density, but seeks would be slower. If you read the article that is the exact opposite of what they want.

Does anyone else agree? SSD is cheap enough now to use it for caching in front of HDDs, even if you use it nowhere else. Google has special requirements which involve minimum cost, because their architecture depends on multitudes of nodes.

Re:Form Factor not "Format"

[Anonymous Coward]

Depending on the process, MTBF (mean time between failure) for Flash is not as high are what you've experienced in your laptop SSD. Your unit falls outside the mean.

Re:Form Factor not "Format"

[drinkypoo]

Um, it hasn't been "rust" in decades. That's as condescending as saying SSD's are lumps of sand...

Poetic license. Besides, spinning is the really relevant part.

Re: Form Factor not "Format"

[Anonymous Coward]

500x4k = 2MBps? Something is not quite right with your model.

Re:Form Factor not "Format"

[Anonymous Coward]

SSDs win out over 15k drives for IOPS per watt of power. You can either setup a huge array of 15k RPM SAS drives that are all short-stroked, or you can simply use a pair of 1TB SSDs running in RAID-1. The SSDs are going to still crush the SAS array on cost, power used, cooling required and throughput and IOPS.

Even for a 10TB array, running say RAID-6, you're looking at 13 SSDs at a cost of about $600-$900 each. A 15k SAS array with similar performance would be at least 50 spindles, eat an order of magnitude more power and cost about the same as the SSDs.

The days of 15k SAS drives are over.

Re:Form Factor not "Format"

[Anonymous Coward]

You should never use RAID5 in production. You either want to run RAID6 or RAID 1+0. The RAID6 can handle a double drive failure, while the RAID 1+0 offers much faster rebuild speeds when a drive fails.

For the truly paranoid, there is 3-way RAID-1 (three drives, mirrored) which can also handle a double drive failure.

Re:Form Factor not "Format"

[Anonymous Coward]

Samsung PM1633a.

It uses 3D NAND and will probably be very expensive at first, but it's definitely possible to get that amount of SSD storage into a 2.5" enterprise form factor (probably 20-25mm, not 9-12mm).

Re: Form Factor not "Format"

[cthulhu11]

Truly only douchbags ever used "vaxen".

Re: Form Factor not "Format"

[cthulhu11]

Will it? Or will the movement to NVMe render it moot for drives?

Re:Form Factor not "Format"

[rthille]

My first HDD was $2000, 330 MB, and full-height 5.25"

But my college house-mate had an external 5MB, about the size of 4 shoeboxes, and sounded like a vacuum.

Re:Form Factor not "Format"

[lsatenstein]

The world will probably keep using spinning rust until purchase price (not TCO) on SSDs is lower. I wouldn't be surprised if makers went back to 5.25 x half height, and low spindle speeds. It would still permit large throughput with high density, but seeks would be slower. Not a big deal with enough caching in front of them, and/or with enough disks in an array. As SSDs approach HDD price, they will take up more of the workloads that actually have to be fast anyway.

SSDs for manufacturers are now below 1.8 times the price of hard-disks. And SSDs allow for thinner lighter laptops, desktops and even some tablets. There are some newer technologies coming for SSDs, to make it even cheaper than HDs. (mass production, no motors, no read/write heads, etc.)

Re:Form Factor not "Format"

[lsatenstein]

That's odd, because my laptop's SSD is four years old and still has plenty of usable life left - and it's from a middle-line vendor, from the early SATA3 days, so it's not even a particularly good SSD. The hard drive in the same laptop (dual-bay) is actually reporting as closer to failure. Maybe that's because it's a laptop, so it suffers more vibration and temperature variation, which is harder on hard drives than solid-state.

And the rest of your bitching seems to be based more on shoddy cloud hosts than SSDs, or on badly-configured servers. "SSDs are too fast, they bring down the entire system by filling up RAM"... wouldn't that be true of hard drives as well, IF they could transfer data that quickly?

In winter, if I left the laptop in the car overnight, the SSD works immediately on power-on and at -20C. The hard disk won't spin at that same temperature.

Re:Form Factor not "Format"

[Agripa]

Quantum tried that with their Bigfoot series of drives when the shift to the 3.5" format made the surplus infrastructure for 5.25" drives inexpensive and it still was not economical. If you did it now, you would have to recreate the infrastructure to build the parts for the 5.25" drives.

Re:Form Factor not "Format"

[niftymitch]

You can? Where? Biggest I can find in 2.5" is 2TB, so where are you finding 16TB SSD drives?

Toshiba has a 3TB drive for laptops.
http://www.kitguru.net/compone...
This comment makes Google's point: "..Since the hard drive employs so many platters, it is 15mm thick, which means that it will not fit into the vast majority of laptops. According to Toshiba, the hard drive is designed for “personal external storage and space-constrained needs”....

In a custom machine room today a "machine" is a thin'ish PWB + processor and memory
where the limits of thickness are bound by the cooling system on the CPU and the stack
of RAM.
This tells me that the drive could be thicker by a lot and still run cool enough.
Once spinning the power of a drive is almost unaffected by the number of platters.
A smaller diameter could spin up quicker with less power.

The SSD flash based form factor is cooling limited but is "simply" chips on PWB
and a Google sized farm could justify an in house build to match space and cooling
needs.
On the cooling side constant airflow with no paths that are "too easy" is important
and it makes sense to replace air dams with functionality. Thus thicker devices
where thin is the current choice makes sense. Spinning media has thermal qualities
that packed Flash, DRAM and friends to not.

My thought on laptops is that thin and light is no longer an interesting feature to me.
Keyboards have suffered to no end. Internal volume for a second drive or additional
battery has approached ridiculous. Hanging an external "thing" on laptops seems
way too common and is simply lame. Heck some twit just patented a computer
that is reconfigurable and made up of parts. The industry has moved this direction
on and off for +40 years. One 40 year old cartoon had a tiny modern like laptop and near invisible
wire to a 3 ton lump of hardware and battery. How is this different from some cell phone
designs.

But laptops are a digression to the needs of Google. Me I would be happy with a laptop that could accommodate
a 15 mm. drive. As long as the screen is not so heavy that the whole thing needs a tripod to stand up like the convertible
touchscreen things that might appear to be popular because they are so strongly marketed.

Re:Form Factor not "Format"

[kimvette]

Home server += "production" but yes I am aware of the advantages of RAID6. RAID5 + hot spare will do. :)

Re:Form Factor not "Format"

[Coren22]

The parent to my comment was stating that there were:

You can buy a 4TB 2.5" 7mm z-height SSD right now. Or a 16TB 15mm.

I can't say that I have ever heard of a SSD as large as 4 or 16 TB. You would be talking about multiple SSDs to achieve that kind of size. You are linking to a spinning rust drive, not a SSD, and even it isn't 4 or 16 TB.

Re: Form Factor not "Format"

[8086]

It doesn't matter, really. OP was saying that the interface will be a bottleneck for these drives and I was trying to say that it won't be.

Re:Form Factor not "Format"

[gbjbaanb]

absolutely,Google wants you to pay for all the infrastructure society provides without paying their share in tax, and now they want you to pay for the development of better seek HDDs for their datacentres too.

Global corporatism mate, its not for your benefit.

Too late?

[Anonymous Coward]

I just wonder if, by the time they agree on this (if they do) the price of SSDs will have dropped enough so that they can be used instead? Storage-wise they are already there, and then some.

Re:Too late?

[JoeMerchant]

I just wonder if, by the time they agree on this (if they do) the price of SSDs will have dropped enough so that they can be used instead? Storage-wise they are already there, and then some.

The point is to keep spinning platters cost-competitive with SSDs - a taller, smaller form factor would increase performance and reduce TCO... I'm thinking they're looking at something like lots of 1.8" platters stacked 4" high, they can spin faster, have faster seek times, and package multiple TB per unit, and I think the longer single bearing should be a more favorable geometry than the ultra-thin notebook compatible drives that have been developed for the last 10 years. It will be slower than SSD, but the power performance (which is the key to TCO) should remain competitive with SSDs for a long time to come. Also, presumably, if this takes off it would be datacenter focused, so longevity (again, TCO focus) should also be "baked into" the design in favor of lower retail price.

Re:Too late?

[Khyber]

The power performance will NEVER COME CLOSE to being as low as an SSD.

Even my first-gen SSDs use far less power than a regular laptop drive. Taller drive geometry = more power to spin the spindle. You do know what an ?INDUCTIVE LOAD is, right? If not, protip for you: The amount of power you use to spin up those platters alone is all the power I need to find and transfer data from my SSD. And that's done before your drive heads even begin moving!

Re:Too late?

[Anonymous Coward]

Spin up? For data centers drives never stop spinning until they die. Data transfer is constant writes, a weak point of SSD.

Re:Too late?

[Anonymous Coward]

Data transfer is constant writes, a weak point of SSD.

A claimed weak point of SSD.
So for all failures of SSDs I've heard of have been driver issues, not wear. (Never had an SSD fail on my myself, but I've only had one OCZ.)
The theoretical values says that the bus isn't fast enough to make wear an issue for SSDs during the lifespan of a spinning disk.

Re:Too late?

[Anonymous Coward]

Rotational inertia is smaller on platters which have a smaller radius, by a factor of R^2/r^2. It's larger on more platters by a factor of N/n. The same constants of proportionality also apply to data storage density. So actually, taller thinner geometry ~= THE SAME rotational inertia so for the same RPM so it takes THE SAME energy to spin the spindle, given equal storage capacities. (Actually it's a *little* more byte for byte, because the spindle doesn't store data.)

BTW, do you know what an inductive load is? Because that has nothing to do with how much energy it takes to start a disc platter turning.

Re:Too late?

[Anonymous Coward]

SSD doesn't hard fail by wear, only reduces size. The SSD could last 20 years and still only have half the sectors marked bad.

Re:Too late?

[castionsosa]

Some drive arrays power down hot spares to save energy (and periodically powering them on to run a drive check). Even though it isn't much energy, keeping platters spinning does take some, and then the HVAC system has to deal with the heat from the platters.

The thing about SSD is that it can store data using zero watts. For dense data centers where power management is essential, this ability is critical, even if SSDs cost more, for saving wattage and showing off a "green" data center. Since SSDs are denser than HDDs, are more shock resistant, and -tend- to have a better MTBF, once cost comes down, there may be no need to bother with a new HDD form factor at all.

Of course, this doesn't say that SSDs are perfect. When they lose data, it tends to be gone for good, as opposed to HDDs where the magnetic domains are present, and can be accessed somehow, barring physical damage from head crashes.

Re:Too late?

[I4ko]

Yet 2 out of my 5 SSDs have failed, and those are reputable brands. The first one developed memory hole that the controller does not see, it is just completely unable to read several sectors (a 64kb piece) and responds with seek error, and has made no attempts at remapping them, the other lost its capacity and is showing like 20mb drive (some kind of firmware update mode), which also has all of it's sectors unreadable.

Re:Too late?

[Anonymous Coward]

Just the cost to manufacture this doesn't make sense given the exponential decrease in cost of SSD. Tape is the more effective 'high capacity low performance option,' and disk will never come close. Meanwhile NAND become easier to produce, and more widely produced. It's surprising that Google would suggest such a course of action.

Re:Too late?

[Tharkkun]

Yet 2 out of my 5 SSDs have failed, and those are reputable brands. The first one developed memory hole that the controller does not see, it is just completely unable to read several sectors (a 64kb piece) and responds with seek error, and has made no attempts at remapping them, the other lost its capacity and is showing like 20mb drive (some kind of firmware update mode), which also has all of it's sectors unreadable.

Some of the 1st generation had issues. But spindle drives still continue to fail far more often than spindle

Re:Too late?

[DutchUncle]

Umm . . . a taller drive geometry in a single package with a single motor (rather than multiple packages) should take less *net* power to spin for the same amount of storage. Make the radius smaller (fewer tracks, less seek time) should lead to less rotational momentum and even less net power. If we're making special data-center drives in vibration mounts, maybe we can talk about magnetic frictionless bearings or other mechanical enhancements. OF COURSE an SSD will always be more efficient - but the cost will still be higher.

Re:Too late?

[toddestan]

My first SSD, which I bought in 2009 as an early adopter, started stuttering horribly within a year. I ended up ripping it out and went back to a HDD (at a significant performance increase over the SSD at that point), which I only replaced this year. I ran some secure wipe on the SSD that was supposed to restore performance. I let it sit on a shelf for a year anyway, and when I tried to finally use it for something also it only lasted a few days before corrupting everything on it. I'd secure wipe it, and it would be fine again for a bit, then shit itself all over again. It finally stopped responding. Other SSDs were not nearly as interesting, as they would just suddenly stop responding. These were all reputable brands too (Samsung, Intel, Crucial). The only brand that hasn't given me trouble is Sandisk.

The can -ropose what they want

[Anonymous Coward]

But they need to convince someone that there is an economic case for making the fucking thing

Re:The can -ropose what they want

[JoeMerchant]

Google$ - the economic case for making almost anything.

Re:The can -ropose what they want

[Anonymous Coward]

True, they are literally the epitome of why antitrust legislation exists, but it's so far past too late that they essentially write their own legislation now.

Re:The can -ropose what they want

[arglebargle_xiv]

That was my thinking as well, we've already got HTTP4Google (a.k.a. "HTTP 2.0"), TLS4Google (a.k.a "TLS 1.3"), Google phones (Android), Google laptops (Chromebooks), they're proposing Google-optimised hard drives, why don't they get their own planet where they can dictate everything the way they want it.

Re:The can -ropose what they want

[castionsosa]

This can be doable, because this is intended for servers, so if drive companies could make the platters, server companies make the hot-pluggable bays, and third party makers make enclosures so the drives can be used as external drives.

I can see a drive using 2.5" platters (so less re-engineering would have to be done by the HDD makers, as opposed to a different diameter), stacking a good number of them in an enclosure. This would allow for some decent sustained I/O, since it could read/write from all those sides at once. Add HAMR and SMR, and these would be quite useful for low tier, read-mostly storage. It would be expected that there would be a very large I/O cache, or an array of SSDs to handle the random reads/writes, because of the nature of these drives.

MTBF, bit rot, and redundancy would be paramount, so hopefully these drives could be designed from the ground up to be more reliable, be it a larger bad sector relocation table, more ECC/parity, drive heads able to take more shock, or so on.

For the consumer, perhaps combine the physical drive with some SSD, so that the flash part works as a landing zone for data, minimizing the amount of random I/O the actual spinning platters have to deal with.

Re:The can -ropose what they want

[shredwurzel]

They could call it "Google Earth"... oh wait...

Re:The can -ropose what they want

[Anonymous Coward]

why don't they get their own planet where they can dictate everything the way they want it.

Because Europe protests against eradicating all privacy and the USA really don't want to go green.

Re:The can -ropose what they want

[Martin Blank]

HTTP protocol improvement was on a slow track until Google introduced SPDY, and then that was used as the framework for HTTP/2. Note that Google didn't get everything it wanted--SPDY requires encryption, while HTTP/2 technically does not (though no major client is implementing it without requiring encryption). It also took input from many industry sources; Google may have been the motivator, but was not remotely the final decider.

I'm not sure about the claim behind TLSv1.3, but so what if Google got it started? The independent analyses I've seen suggest that it has substantial improvements (DH required, shorter handshake, some other things) over TLSv1.2, so why not go with it, especially since it's being done as an open standard?

At this point, we have so much legacy protocol cruft with everyone afraid to move forward for fear of breaking things that I'm happy to have Google push the boundaries. They don't win on everything (look at how many projects they've shut down because they don't get the hoped-for uptake), but they're willing to do things that improve the Internet as a whole.

Now, if they could just start working on a replacement for SMTP, maybe we could get some real improvements.

Re: The can -ropose what they want

[Anonymous Coward]

They had. Google Wave.

2.5"

[Anonymous Coward]

I find 2.5" preferable for desktops/laptops. But for server farms — I doubt it matters at all what size is practical for PCs.

Re:2.5"

[Anonymous Coward]

Imagine how much data you could store on a drive with a platter size that just barely fits in a 1U... Sadly, that probably isn't feasible.

Re:2.5"

[fisted]

Probably around the same as you could store in a 1U enclosure filled with 2.5" drives. Okay, a bit less, but at least one could seek in reasonable time.

Go for it! Bring back full height 5 1/4" drives

[Anne Thwacks]

but keep the compatible mounting holes!

Multi-platter was always a good idea, I assume it stopped in a desperate attempt to cut costs.

8" hard drives often had 4 or even 8 double sided platters - and SCSI interfaces! Early 5.25" drives often had two, double sided platters. They desperately needed to access more data with less head movement because they had quite low areal bit density and used floppy-derived stepper motors for head positioning!

Re:Go for it! Bring back full height 5 1/4" drives

[technosaurus]

I had several quantum bigfoot drives that were a really good value at the time and worked great with all of the otherwise useless extra 5.25" slots.

Re:Go for it! Bring back full height 5 1/4" drives

[mrchaotica]

Multi-platter was always a good idea, I assume it stopped in a desperate attempt to cut costs.

Wait, what? Last time I opened up a dead 3.5" hard drive (which was only a few years ago) it had either three or four platters. Are you saying they typically only have one now?

But yes, I agree that if they want taller drives, 5 1/4" full height would be a good form factor. Maybe even not with 5" platters! If they want quicker speeds, they could maybe put four separate spindles of the platters from 2.5" drives inside the same box.

Re:Go for it! Bring back full height 5 1/4" drives

[jeffb (2.718)]

It sounds like you think that manufacturers have stopped making multi-platter drives. That's not true. Seagate and WD both use seven platters in their highest-capacity (10TB, standard-height) drives. The linked article further states that they use seven platters "instead of the usual six".

I don't know how prevalent single-platter drives are today, but multi-platter drives certainly haven't disappeared.

Re:Go for it! Bring back full height 5 1/4" drives

[mrchaotica]

Actually, I just realized they could do even better: put four spindles of 2.5" platters in a 5.25" case, then put a fifth spindle in the center with the platters vertically offset to interleave with the others!

Re:Go for it! Bring back full height 5 1/4" drives

[kschendel]

Cost may have been a driver, but another driver was the lighter and therefore potentially faster head positioner assembly. Lighter positioners allow you to either move them faster or use less power, or some of each.

Re:Go for it! Bring back full height 5 1/4" drives

[Anonymous Coward]

our first year failure rate was over 40% on those...so, not good value. We had stacks of them, Micro stopped taking them back, they accepted a paper form in lieu.

Re:Go for it! Bring back full height 5 1/4" drives

[andphi]

It depends on the age of the drive, the manufacturer, and capacity, I think. Mostly capacity, probably. Most of the SATA drives I've taken apart recently had only one platter. A few have had two or three.

Re:Go for it! Bring back full height 5 1/4" drives

[omnichad]

Parent poster is probably just buying small drives. Economies of scale says it's cheaper to manufacture one platter density and just vary the number of platters. So most 500GB drives are single-platter now (either 500GB platter or defected 1TB platter). Most newer drives are probably 1TB platter. So anyone who avoids 3TB drives because they're "unreliable" is missing the point.

Re:Go for it! Bring back full height 5 1/4" drives

[Jahoda]

I'm not really sure what you're talking about. Hard Disks have been and continue to be multi-platter. The latest 10 TB Seagate HE drive has 7 platters and 14 heads, and this is only their playing "catch up" with Hitachi/WD. http://arstechnica.com/gadgets/2016/01/seagate-unveils-its-own-10tb-helium-filled-hard-drive/

Re:Go for it! Bring back full height 5 1/4" drives

[threephaseboy]

It depends on the capacity, generally the lower capacity ones will have fewer platters or even a single head
(pdf) 250GB-3TB: 500-1TB per disk, 1-3 disks

(pdf) 3TB-8TB: 1.3TB per disk, 3-6 disks

So a cheap 1TB drive may have a single platter, but they add more as needed.

Re:Go for it! Bring back full height 5 1/4" drives

[ChrisMaple]

Where are you going to put the head actuator arm for the center spindle?

Re:Go for it! Bring back full height 5 1/4" drives

[mrchaotica]

Nowhere, because I'm stupid and didn't think of that problem until after hitting "submit."

Re:Go for it! Bring back full height 5 1/4" drives

[jandrese]

You are probably taking apart drives that were purchased by cheapskates. What usually happens is you see a manufacturer announce a new drive line with 2TB, 4TB, and 6TB capacities and what they do is sell you a drive with either 1, 2, or 3 platters. If your purchasing guy is looking to cut costs he will only buy the lowest end drive in the line.

Re:Go for it! Bring back full height 5 1/4" drives

[Anonymous Coward]

Single platter is pretty much just used for low-end consumer stuff and high-end performance stuff.

Re:Go for it! Bring back full height 5 1/4" drives

[toddestan]

Typically the smallest drives are only one platter. The larger drives will have more platters to get the higher capacity. For example, they get 1TB per platter, so the 4TB model will have 4 platters, the 2TB will have 2, and the 1TB model only one. The 500GB model will use one side of one platter. Counting platters is one way to tell how advanced the drive was when it was made (well, other than looking at the date code :). A 4-5 platter drive at a capacity was probably one of the first drives of that size on the market. If there's only one platter, it was probably one of the last drives of that size made.

Re:Go for it! Bring back full height 5 1/4" drives

[toddestan]

If I had to guess, the single platter drives outnumber all the other configurations. Virtually all the 500GB-1TB drives are single platter drives now, and that's what you'll find in all the low end computers.

Re:Go for it! Bring back full height 5 1/4" drives

[OffTheWallSoccer]

It was a fun visualization, though!

Bring back 10 inch disks

[Anonymous Coward]

Bring back these removable 10 inch disk packs - 1970s tech
http://www.nf6x.net/2014/03/da...

Multiple heads

[chriswaco]

Multiple heads on each side of the platter might be a better solution, one for the inner part and one for the outer.

Re:Multiple heads

[jones_supa]

I also have been curious as to why HDDs have never introduced multiple heads. I have visioned it being two heads, placed on opposite sides. Combined with smart NCQ, it could be quite sweet.

Re:Multiple heads

[Andrew Lindh]

This has been done before.... Both outside/middle dual heads and dual independent actuators on each side. Multi heads can increase performance, but cost space, power, and money. Also more parts = lower MTBF. They don't increase storage density. If you want performance use SSD.

http://www.tomshardware.com/ne...

Re:Multiple heads

[nojayuk]

There were SCSI drives with four head actuators, one in each corner of the drive casing. They were treated as four separate drives logically and used to speed up reads on a "first to deliver the requested block" basis. They were horrendously expensive and it turned out to be very difficult to optimise the read process to gain the desired perfomance boost.

Re:Multiple heads

[castionsosa]

There was one drive maker which actually did this. They had two drive platters at opposite ends, each independent of the other, and either could fail, letting the other completely take over. I've wondered why this isn't more commonplace, perhaps a drive form factor with four heads, all active/active and can handle a head array failing (perhaps lighting up SMART.) This wouldn't just allow for four times the I/O, but allow four different threads to write at the same time, which is useful for virtualization, although these days, virtualization should just go to SSD or a large I/O buffer due to all the random reads/writes.

Re:Multiple heads

[Anonymous Coward]

I also have been curious as to why HDDs have never introduced multiple heads. I have visioned it being two heads, placed on opposite sides. Combined with smart NCQ, it could be quite sweet.

Was done by Conner Peripherals ages ago. The problem was that it had double the moving parts and double the electronics so cost twice as much as a regular drive. Not to mention that the additional heads & electronics increased the heat output.

Re:Multiple heads

[omnichad]

This wouldn't just allow for four times the I/O, but allow four different threads to write at the same time

Or allow writes to not block reads.

Re:Multiple heads

[Anonymous Coward]

i believe multiple heads per recording surface was researched years ago but the additional high precision moving parts such as read/write heads a much greater liability in terms of reliability for the modest performance gains it would give.

when i first read about the new dual platter 'slim' form factor notebook drives i started thinking why don't we have half height 3.5in drives, like the good ol' days, to squeeze more platters, more easily, into a desktop hard drive.... and now this comes out.

Re:Multiple heads

[tlhIngan]

There was one drive maker which actually did this. They had two drive platters at opposite ends, each independent of the other, and either could fail, letting the other completely take over. I've wondered why this isn't more commonplace, perhaps a drive form factor with four heads, all active/active and can handle a head array failing (perhaps lighting up SMART.) This wouldn't just allow for four times the I/O, but allow four different threads to write at the same time, which is useful for virtualization, although these days, virtualization should just go to SSD or a large I/O buffer due to all the random reads/writes.

There was a drive that had two actuators that could access the entire platter. That was the design, but I think in the end the complexity of multiple heads accessing the same sector was problematic because of the ordering of the operations (i.e., one head could write data to the sector that the other head was reading - if you didn't catch this, you would corrupt the data).

Plus, double the heads doubles the chance of a head crash.

Re:Multiple heads

[Megol]

IIRC (was a while since I last saw an answer to that question) it wouldn't make economic sense as making two HDDs would cost about the same and would perform as good or better. Also the combined unit would be more sensitive to a headcrash.

Re:Multiple heads

[Cesare Ferrari]

I had a Fujitsu Eagle from the 80s which used multiple heads per side. The drive was used on a PDP-11, was 19 inch rack mount, and had a perspex cover, so you could watch the heads seeking when the drive was in use.

Re:Multiple heads

[iggymanz]

why not just a immovable heads, head per track (heads staggered of course as head much bigger than track but sufficient flux in center pushes the domains over)

Re:Multiple heads

[cnettel]

There was one drive maker which actually did this. They had two drive platters at opposite ends, each independent of the other, and either could fail, letting the other completely take over. I've wondered why this isn't more commonplace, perhaps a drive form factor with four heads, all active/active and can handle a head array failing (perhaps lighting up SMART.) This wouldn't just allow for four times the I/O, but allow four different threads to write at the same time, which is useful for virtualization, although these days, virtualization should just go to SSD or a large I/O buffer due to all the random reads/writes.

There was a drive that had two actuators that could access the entire platter. That was the design, but I think in the end the complexity of multiple heads accessing the same sector was problematic because of the ordering of the operations (i.e., one head could write data to the sector that the other head was reading - if you didn't catch this, you would corrupt the data).

Plus, double the heads doubles the chance of a head crash.

It seems like NCQ, write and read caches (sometimes with flash hybrid modes) etc in current drives would bring enough complexity that additional physical heads would also be reasonable to implement. The abstraction in the drive firmware is much thicker these days.

Re:Multiple heads

[dfsmith]

A few years ago, drives were about 40,000 tracks per inch. Each head costs about $2. Any more questions?

Re:Multiple heads

[jones_supa]

I didn't mean two HDDs but two heads.

Re:Multiple heads

[iggymanz]

typical, you are envisioning the "head" as you know it having functional fixedness, I am not. You could not even comprehend the layout which would NOT have 40,000 heads per inch next to each other

So how will this limit desktop users?

[Torp]

I have a feeling that in a few years we'll be left with just expensive SSDs and even more expensive "datacenter" drives.

Re:So how will this limit desktop users?

[hackertourist]

Expensive? SSD prices have been dropping like a rock for several years, getting closer to HDD by the month.

Re:So how will this limit desktop users?

[Anonymous Coward]

The data center centric drives Google is proposing will not be more expensive. They make a few suggestions about what to do to improve performance for data center uses cases. But what it basically comes down to is this: Given the choice between good, fast, and cheap Google will take fast a cheap.

They have to cycle so many drives out just to keep up with expanding storage capacity that drive failures aren't a major concern. They'd rather have faster drives that are cheap to replace because they all need to be replaced even when they don't fail.

If it does have any effect on consumer hard drives it will be an increase in drive failure rates.

Re:So how will this limit desktop users?

[Anonymous Coward]

Desktop? Is that some new form of tablet?

Re:So how will this limit desktop users?

[squiggleslash]

Just bought a half terabyte laptop SSD for about $150, which was roughly what a 2.5" magnetic drive with equivalent capacity cost me three or four years ago, and only about twice the current 2.5" magnetic drive cost. I know, I know, still not comparable to 3.5" costs, but it gives you some idea of how quickly the prices are plummetting.

Re:So how will this limit desktop users?

[threephaseboy]

roughly what a 2.5" magnetic drive with equivalent capacity cost me three or four years ago

Try 7 years ago.

Re:So how will this limit desktop users?

[Anne Thwacks]

A table with an A0 size screen? I want one!

Why no use existing form factors?

[GuB-42]

There are other form factors other than the typical low profile 3.5".
In particular there is the "half-size" thickness, witch is the thickness of 5.25" bays. It was a rather common form factor for 3.5" SCSI drives.

Re:Why no use existing form factors?

[drinkypoo]

In particular there is the "half-size" thickness, witch is the thickness of 5.25" bays. It was a rather common form factor for 3.5" SCSI drives.

It was popular from the end of the ST-506 era up into the early days of ultra SCSI. However, the benefit of making a taller drive is being able to stack in more platters, which means you also need more heads and so on. Instead, they improved areal density, so that they could make the disks shorter. Now we're all married to the 3.5x1" format because of drive sleds and so on. We are, however, free to use 5.25" storage devices of whatever height we want, whether that's 1", half-height, or full-height. Those are still commonly managed with rails rather than sleds.

Re: Why no use existing form factors?

[bill_mcgonigle]

Google could build a full-height 5.25" 'sled' that had a logic controller on it and a slide-out tray that would house six 'data cubes: each containing platters and heads that could be plugged in or out as they failed or needed upgrades. Replicating the logic 6x over is silly given today's CPU's. Frankly these things need to be SAS for compatibility but really just run PCIe to the sled and skip the discrete controller too, to get costs down more.

I'd buy such things if they were on the market.

Re:Why no use existing form factors?

[Anonymous Coward]

DLT tapes were that size, they had the grooves in them so you could use 2 3.5 bays to hold them. I'm sure a modern case would still hold those, so it could hold a double height 3.5 drive as well.

Urs Hölzle - Moron

[Khyber]

"In a new research paper the VP of Infrastructure at Google argues for hard drive manufacturers and data center provisioners to consider revisions to the current 3.5" form-factor in favour of taller, multi-platter form factors"

Google needs to fire this idiot. First off, we moved away from the larger drives BECAUSE THEY REQUIRE MORE POWER TO OPERATE AND GENERATE FUCKTONS OF HEAT. Secondly, SSDs are at and beyond the capabilities of spinning rust under a layer of vapor-deposited platinum. Third, SSDs are much smaller - even the typical 2.5" laptop drives are mostly empty space. Fourth, the energy required to manufacture an SSD is about 25% of what is required to make a comparable HDD.

Can someone tell me what 'University' this 'VP of Infrastructure' attended so I know not to send my children to such a moronic 'Institution of Education?'

Re:Urs Hölzle - Moron

[Anonymous Coward]

He probably attended one that had economics as part of the curriculum. You know, SSDs are way too expensive for what he needs.So it doesn't make sense to buy millions of SSDs when spinning drives are much cheaper. He has fucktons of power available. Some of their data centers have hydro power right next to them.

Re:Urs Hölzle - Moron

[pr0nbot]

It's true, Google are renowned for hiring only morons. I'm told that at the interview people are asked stuff like, is MongoDB web scale? And I'm sure they only promote the total chumps to VP. All he had to do was post an "Ask Slashdot" and you'd have no doubt politely schooled him. What a wankpuffin he must be!

Re:Urs Hölzle - Moron

[operagost]

Old 5.25" half and full height (and 3.5" in ye olde 1.6" form factor) did use a lot of power and generate a lot of heat. But besides modern power supplies being more efficient, he's asking for smaller platters stacked higher, not the large circumference drives of old. The smaller circumference is to reduce seek times, and that reduction of inertia is sure to address power and heat concerns at the same time. I'd be curious as to whether one of these tall, skinny drives would really generate more waste heat than the two or three smaller form factor drives (with two or three more motors and logic boards) it would take to provide the equivalent capacity.

Re:Urs Hölzle - Moron

[Khyber]

I've worked for Google. They used to have a serious vetting process. Now it's more of who you know than what you know.

Re:Urs Hölzle - Moron

[Megol]

Oh tell me vise one, how many years experience in HDD manufacturing do you have?

Calling one of the most advanced technologies existing in this world "spinning rust" tells me 0 knowledge which would make you a loud-mouthed know-it-all asshole. The lack of insight in that HDDs is a leading storage technology with several advantages (including economics) over SDDs just reinforce that.

But if your clueless rant makes you feel better that you haven't (and will not) reach a comparable position then I guess it is worth something. Just not for others.
 

Multiple new form factors would be better

[Letophoro]

Realistically, there should be several size formats for different purposes and market segments. Google wants a new format for their purposes. That doesn't mean that it would necessarily be good for the needs of someone else.

I personally wouldn't mind seeing the return of the 5" HDD.
Given the more than doubling of area (capacity) in each platter by going from 3.5" to 5", I could live with higher seek times to have a 16TB HDD taking up one of the 5" bays in my PC case.

Re:Multiple new form factors would be better

[Letophoro]

Dagnabit! I meant 5-1/4" form factor.

Re:Multiple new form factors would be better

[jofas]

You would never see this technology in your PC in the same way you don't' currently have any 15K SAS drives or Fiber Channel host adapters.

Re:Multiple new form factors would be better

[KGIII]

While you're correct in my case, with that specifically, you'd be surprised at what some of us have in our home server closets. I have two racks in mine but nothing fancy like that. I even have a small, but older now, blade server from HP. I have the conduit for fiber but haven't put it in yet. I'd do it but I don't know how to do the connections and splicing. I've played around with it on a friend's work equipment but I'm not actually any good at it.

Re:Multiple new form factors would be better

[Anonymous Coward]

Most Google servers use mostly-custom hardware with cheap SATA disks.

2.5" 4X drives

[wren337]

Surprised they haven't just gone with 2X or 4X height 2.5" drives. Same connectors, same platters, easy retrofit. You just need a different bracket.

Re:2.5" 4X drives

[0100010001010011]

I'm surprised that they haven't just done away with the 'hard drive' as is. SSDs are just a bunch of chips. I'm thinking of a 1U server that is just a board populated with chips, a fiber interface and a powersupply. Treat the 1U server as a single unit.

When you start to add up hard drive casing, interface connectors, etc you end up wasting a lot of space for no reason. For the home user that only has 1-2 drives they make sense but for someone like Google that may have thousands of drives just jump up to the next standard unit and make that the 'storage device'.

Re:2.5" 4X drives

[Anonymous Coward]

I'm surprised that they haven't just done away with the 'hard drive' as is.

Once you have a budget and money isn't infinite, you aren't surprised at all. That's why Google wants thinker drives. Cheaper storage. What's funny is the article says the size came from floppy drives. It did, from half-height floppy drives. Just go back to full height drives for lots more platters.

Re:2.5" 4X drives

[DerekLyons]

Surprised they haven't just gone with 2X or 4X height 2.5" drives. Same connectors, same platters, easy retrofit. You just need a different bracket.

I'm not (surprised). The case size of a PC tower has been trending steadily downwards for the better part of a decade. There's not room for an additional drive of that size in the common consumer tower anymore.

apk

[Anonymous Coward]

apk has been saying this for years!!! Turns out this guy is brilliant, not a lunatic. -apk

Re:apk

[Anonymous Coward]

What's an apk? A new package manager?

Re:apk

[Anonymous Coward]

I thought you got banned APK? Welcome back dude!

Re:apk

[Anonymous Coward]

Android application package

Re:apk

[U2xhc2hkb3QgU3Vja3M]

I thought it was a HOSTS file with a built-in ad server.

Eric Brewer = Moron

[Khyber]

Hey, Eric? Google needs to fire you. You're too stupid to be any visionary of the future.

First, spinning rust drives eat fucktons of power and generate fucktons of heat. This is a NO-GO for any datacenter wanting LEED or any other 'green' certification. Second, SSDs meet and beat platter drives in every way conceivable. Third, the cost of manufacturing an SSD is about 25% that of manufacturing a platter HDD. Fourth, even as drive sizes shrink, drive capacities are still increasing.

Someone please tell me where Eric Brewer attended university, so I may not send my children there to receive such a poor education.

Re:Eric Brewer = Moron

[110010001000]

Wow. That was harsh. Too bad you are incorrect. The cost of manufacturing a similar size SSD as a typical datacenter size drive is much much much more. Plus you don't need SSD to get LEED certification. Ridiculous.

Re:Eric Brewer = Moron

[yodleboy]

Could be that Google has some inside information that leads them to believe that prices on SSD will not be dropping to acceptable levels any time soon, despite what SSD boosters would have us all believe. If they are proposing something like this, they must have some inkling that spinning platters have a great deal of life left.

Re:Eric Brewer = Moron

[Anonymous Coward]

That's one of the weakest papers I have seen from Google.

1) new form factor - this will happen as soon as there are no more 3.5 inch bays outside the servers/data center (any day now, see 3TB 2.5 inch).

2) remove cache from drives - the cache is already getting smaller and smaller as a ratio of capacity, and as they point out it is very effective still. They want larger IOPs/blocks (#9), the drive will need to cache at least a few of those to keep the data path SAS/SATA/new free.

3) it seems they don't know the Seagate 8TB SMR, which already combines SMR with PMR (they call it CMR).

4) already solved by read racing in erasure coding.

5) host managed SMR are available.

6) see #4

7) see #4

8 ) no comment , NDA.

9) See #1

10) See #5

Re:Eric Brewer = Moron

[jofas]

+1 for repeating yourself almost exactly. +5 For not being correct. There is more involved in managing data center performance than simple access times or temps. Even with multi-ton cooling, a data center is *still* looking at a lower operating cost to spin hot drives than to use super $$$ high-capacity SSDs. SSDs do not solve any problems for data centers, because individual drive access time is not interesting to a data center. Rather, the performance in $ per KIOPS for the entire array is the real measure of performance.

Re:Eric Brewer = Moron

[Anonymous Coward]

I'm not sure where you learned all this stuff about data centre design ... but I wouldn't send your kids there either.

Hmm, Google's VP of Infrastructure vs some random on the internet. Wonder who knows more about the economics of data centres.

Re:Eric Brewer = Moron

[darthsilun]

the cost of manufacturing an SSD is about 25% that of manufacturing a platter HDD

Really? I think if that were anywhere near true it would be reflected in the cost of SSDs. Do tell, where can I buy a 4TB SSD for $30!

The disk drive market is pretty competitive. I tend to think if SSDs cost 25% of an HDD to make, they'd be selling for a lot less than they are. And with Google's buying power probably even less for them.

Re:Eric Brewer = Moron

[Khyber]

" Plus you don't need SSD to get LEED certification."

No, but every single power saving and low-impact device you use brings you closer to attaining LEED certification. Hard Drives, with platinum (not an easy thing to mine and refine, speaking from personal experience owning two mines here in SoCal) you're ripping the shit out of the environment to get it in most cases.

"The cost of manufacturing a similar size SSD as a typical datacenter size drive is much much much more."

Uhh, what? Do you even know the materials used in an HDD vs an SSD (hint, the spinning rust drive uses expensive shit like platinum and helium.) The BOM alone is 1/4 the cost. It's only expensive because it's sold on performance and a 'new technology' versus HDDs.

Re:Eric Brewer = Moron

[Khyber]

"SSDs do not solve any problems for data centers"

This is patently incorrect and the power bills alone prove it. Go ask any datacenter that switched over to SSD. There's a reason my box uses SSDs (lower requirement for cooling the TWELVE GPUS INSIDE.)

I design these kinds of systems for a living, among other things like owning a mine and mining gems and minerals, designing semiconductors, and much, much more.

"There is more involved in managing data center performance than simple access times or temps."

Most of that unnecessarily complex and useless. Feature creep in the name of "advancement" when nobody ever uses those features excepting maybe 0.000001% of the population.

Re:Eric Brewer = Moron

[Khyber]

"Really? I think if that were anywhere near true it would be reflected in the cost of SSDs."

SSDs command a higher price premium because of A. Marketing B. actual performance and C. perceived 'new technology' to the general mass market, so the prices remain high. This is basic economics 101, man.

Re:Eric Brewer = Moron

[110010001000]

Really? Go build a 8T HDD and a 8T SSD and compare the manufacturing cost. The manufacturing cost of the 8T HD is about $80.

Re:Eric Brewer = Moron

[Anonymous Coward]

After you've finished circlejerking over SSDs, you might realise that for businesses, they aren't all that over a HDD. The price and the performance are good enough that you can just pick up a box of disks and be reasonably happy for the next 6 years, by which point it's time to do a hardware refresh and you'll end up moving all your data to Azure.

I'm sure the same can be said for datacenters too, as this thing called money isn't unlimited.

Re:Eric Brewer = Moron

[Torp]

If you have 12 GPUs in a box, i fail to see how using a SDD or HDD would make any noticeable difference in power and heat...

Re:Eric Brewer = Moron

[ChrisMaple]

Assume a HDD uses 10 watts at $0.08 per kWh. That's $7.00 per year. Allow another $3.00 per year to cool the drive. If the drive needs to last 3 years, the HDD accounts for $30 of electricity over its lifetime. No large capacity SSD is within $30 of a hard drive.

Re:Eric Brewer = Moron

[Megol]

Incorrect is the polite way to express it. Droolingly clueless is a better description of the level of ineptitude displayed.

Re:Eric Brewer = Moron

[ChrisMaple]

A 1 Tbyte SDD sells for about $250. The flash IC's that are used in such a drive cost a total of $500 when bought in quantities of 1000. Even with huge quantity discounts, that doesn't leave much profit margin.

The competing hard drive sells for $50.

The idea that a 1 Tbyte SDD can be manufactured for 25% of $50 = $12.50 is beyond absurd.

Re:Eric Brewer = Moron

[110010001000]

I'm a people person dammit! What the hell is wrong with you guys???

Re:Eric Brewer = Moron

[Pentium100]

$10k for a ~12TB max of storage (if you used RAID0). I paid ~$800 for 6x3TB WD REDs (used in RAID6), getting the same 12TB. According to WD, the drives use 4.1W each when being accessed. That's ~25W total. Let's say the PSU is 80% efficient, so that's 32W from the wall. Electricity costs $0.12/kWh, so for the $9200 I saved by buying HDDs, I can get 76MWh of electricity, which would be enough to run these drives for 270 years.

Hell, for that money, I can pay all my electricity bills for 3 years.

270 years until SSDs pay off is a bit too long in my opinion. Of course, if you need the performance, then there is no way around it, but for me (infrequently accessed archival storage and backups of other servers), HDDs are enough.

Actually I recently built a ~100TB VOD server using a big RAID10 of hard drives with a couple of PCI-E SSDs (Intel P3700) as cache. The client would not have paid for 100TB of SSDs.

Re:Eric Brewer = Moron

[jeffb (2.718)]

Since he keeps emphasizing how hard drives are made with expensive and environmentally-hostile platinum, perhaps he's just comparing the cost per kilo of platinum to the cost per kilo of sand. After all, that's what flash memory is made with, right?

Re:Eric Brewer = Moron

[DutchUncle]

Or else it's just about the sheer quantity of data involved. Even a comparatively small additional cost-per-gigabyte adds up when you're storing all of Youtube plus caching every single website being indexed.

3.5" disks, the 80's are calling...

[Anonymous Coward]

We dont have any 3.5" disks in our data center, all smaller SAS disks.

Spend the money on RAM/SSD and cache everything

[Anonymous Coward]

Are we still talking about hard drives and seek time? In 1999 Jim Gray said "These calculations suggest the simple rule: Cache web pages if there is any chance they will be re-referenced within their lifetime." Rules of Thumb in Data Engineering

Drums!

[kschendel]

Taller, more heads, smaller platter, less seek distance -- the logical end point is the drum! I'm sure we can do better than the FH-1782 today.

Everything old is new again...

Re:Drums!

[vtcodger]

From a programmer's POV, drums were wonderful. Select an address, then read or write. No cylinder/head/sector calculations. No variable transfer rates. You needed better "seek" time, install multiple sets of read/write heads. Unfortunately, they were bulky and cost a LOT.

barking up the wrong tree

[Anonymous Coward]

We are not in nineteen-sixties any more. Instead of rw head on a moving hand, we should have full random-access rw disk placed above the platter. We have wafers, we have photolitography in tens of nanometres of resolution, so why are we pretending that we need to do things like they were when disks were concieved? It is not like anyone is going to remove platters from the housing, so that heads need to retract out of the way. We can have a multitude of rw "faces" per each track, eliminating not only the seek time, but also shortening latency till sought sector begins (or conversely, slowing down angular speed for longer lifetime). Besides, there is possibility of not only random, but also non-blocking parallel access!

Re:barking up the wrong tree

[DidgetMaster]

Someone needs to invent a RW 'bar' that is long enough to go from the spindle to the outmost track. Any point on the bar would be capable of reading or writing. Instead of moving a narrow head on an actuator arm, you just 'activate' (electronically) the part of the bar that is currently over the track you want to access. If cheap enough, you could install a dozen of these bars all around the disk so that you would never need to rotate more than 30 degrees of a circle before a sector you want is under one of the bars. Rotational latency would be very low even on a 2000 rpm drive. Even better would be the ability to activate multiple points on the bar simultaneously so that you could read or write multiple tracks at the same time.

Stupid idea

[Anonymous Coward]

The more platters they put in the more expensive the hard drive is going to get. These things don't exactly have perfect yields, much like a CPU in which not every core on every processor necessarily makes it through manufacturing in a state that is usable, not every platter makes it through manufacturing. Now increase the number of perfect platters you need and all you are doing is driving up the price of one of the most expensive parts. Far better off going with SSD drives since the quantity will help increase production of the drives and bring down the price for everyone. They have already made it clear that storage density is not their concern but seek times are. Typical VP that doesn't understand how things are made just saying: "put in more of them, that'll make it better!"

Horse sense

[TheRealHocusLocus]

The current 3.5" HDD geometry was adopted for historic reasons --- its size inherited from the PC floppy disk.

The form factor of 3.5" floppy drives was decided during the early planning stage of the Great Data Railroad. You can place exactly 16 3.54" (90mm) bare floppy discs side by side within the standard railroad gauge of 4 feet 8.5 inches. For the original 1982 HP single sided format of ~280kB this yields roughly ~4.3mB along every 3.5" of railroad track, or 137 rows along the floor of a a standard 40-foot railroad boxcar without the use of stacking. Thus ~600MB was the capacity of a original single density data railroad car, though it was only only ~1mm in height.

While the floppy disc made data railroads possible, media stacking made them practical. A cylinder of bare floppy media ~10 feet high is roughly 3048 discs, so your standard railroad boxcar held ~1.8TB of floppy storage, in 1982! With an average rail speed of 18mph a single boxcar passes every ~1.5 seconds, which is ~1.2T terabytes or 9200 gigabits per second! By 1998 floppy media storage density had improved ~714-fold, yielding transfer rates of 6568800Gb/s or ~821 TB/s.

So why was floppy data railroad ultimately limited to this 'arbitrary' ~821 TB/s? Northern rail gauge of the US railway based on the English rail system which were based on tramways which used the same jigs used to build wagons whose wheel base was determined by ancient ruts that were left by Roman chariots which were sized to accommodate the width of two horses' asses. As not-quite debunked here.

So the short story is, any chain of decisions regarding technology leads back to some horse's ass.

Re:Horse sense

[Anonymous Coward]

Well played Sir!
You win today's internets...

Re:Horse sense

[Anonymous Coward]

+5 funny, not +5 informative

Re:Horse sense

[jeffb (2.718)]

<slow clap>

Re:Horse sense

[Anonymous Coward]

Ever consider writing material for politicians? You'll have to step up your game a bit though. Need better BS.

file not found

[Spaham]

The research paper is not available. Any pointers ?

Re:file not found

[Anonymous Coward]

Found by typing the filename into this thing called "google":
http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/44830.pdf

For data storage sites why not...

[Rhipf]

something like this?

http://www.101101.io/storageen...

I realize this actual device had terrible data density by today's standards but I'm sure they could do something similar with modern platters.

Another idea...

[Anonymous Coward]

I always believed that the Oval (tm) format had the best reliability and bit density.

CAP === 'excels'

The majority of data is "cool"

[HockeyPuck]

Most of the data within these companies is "cool" meaning it's not actively being accessed. Take for example the massive amount of photos within FB. When was the last time you looked at a photo from 6months ago? If it needs to be accessed frequently, as in it becomes viral, then you move the photo from HDD to SSD. Sure there's 3-4TB SSD coming, but they're still much more expensive $/GB than HDD.

Also, Google's point isn't so much about $/GB but rather that they don't need as much reliability of the drive. Why buy enterprise grade drives when you're making 3+ copies of the data? Consumer grade drives can work for this. If your operational processes are smooth enough, the physical replacement process becomes trivial. Sure there's a lot of them to replace, but it's like replacing books on shelves. So if the cost of a drive can go down even further because it's 99.5 instead of 99.9, even better.

Re:The majority of data is "cool"

[DutchUncle]

Agreed. And scale matters. A tiny difference in cost/byte makes a difference when multiplied by the huge number of bytes.

More platters means less reliable?

[h4ck7h3p14n37]

I would think that adding additional platters would greatly lower the mean time before failure on the drives?

The disk's spindle motor and actuator are shared across platters, but the media and read/write heads are per-platter. With many small platters your seek times would go down, but the odds of a head crash or media failure would be greatly increased.

What's old is new...

[klubar]

Multi r/w heads aren't a new concept. Some of the really old drives had them, and in fact the very original magnetic recording "disks" had a r/w head per track. I think in the trade off of more heads versus faster spinning, faster spinning won out.

I seems that there should be a market for more platters, in a slightly different form factor.

Bring back 5.25" Full Height form factor

[Anonymous Coward]

If my maths is correct, eight 2.5" hard drives will fit into the 5.25" full height form factor.
According to Wikipedia, the hypothetical 5.25"FH hard drive can have 4TB x 8 = 32TB capacity. Maybe even 36TB as extra platters could be squeezed in.
Because it is internally 2.5" platters, it should have very quick seek times.
I guess the problem is the MTBF will drop from 1.2 million hours to only 150,000 hours.

Must maintain backwards compatibility

[nbritton]

Any new solution would have to maintain backwards compatibility. The new standard would have to be ether 3.5" x 2, 3, or 4 bays; or 5.25" x 1, 2, 3, or 4 bays. The industry has 30 years behind existing bay standards, it would take them a long time to change their tooling.

Personally I thought the Sun Fire X4500 (a/k/a thumper) was a very efficient way to maximize storage density

Re:Must maintain backwards compatibility

[iggymanz]

no! they can totally change form factor it as servers mostly are cycled out after five years. those that hang onto them more can scrounge for older drives or cut the metal dividers out of their drive cages. can't let dinosaurs rule the server world for stupid reasons.

does not compute

[Anonymous Coward]

they're using the cheapest consumer disks they can find, but now ask for a form-factor tailored for data centers. This does not compute. Why should such data center for factor disks get such consumer use that they can be had for the same low price?

Of course, consumers love everything that's called industrial quality, so maybe it does work out...

About 20 years too late

[Anonymous Coward]

So Google wants higher performance conventional hard drives? This should have been proposed around 1995. In today's world this has no chance.

You want performance? Get high performance 15,000 RPM drives from Hitachi or Fujitsu or IBM, or whoever makes them now. Just know that all really high performance drives are going SSD.

Current drives are made to optimize a combination of price, performance, reliability, and the standardized drive bays. Google attempts to revise the priorities so that performance ranks higher in the optimization criteria.

The thing is, does anyone care what Google wants? Are they big enough to get custom runs of HDDs made? Maybe, but I doubt it. Seriously, hard drives are a mature technology. And if performance is really the issue you need to abandon mechanical storage and go all-electronic.

I assign this issue to the That Ship Sailed Long Ago category.

Sideways

[alva_edison]

This is not bounded by reality, but just some back of the napkin types stuff.
Let's say you have a 3" platter w/ 1TB capacity. And you can get up to 7 in a 1-inch high 3.5" drive.
That's 7TB.
The spindle is about 1" in diameter, but from looking at the IBM microdrive, it may be possible to reduce that to 0.33"
Next let's shrink the platter to 0.75". Because we're talking single speed, the amount of data is proportional to r (instead of r^2). So it's 0.42/2.5 = 0.168 TB.
The drive is 5.75" deep, Assuming 1.75" for stuff, that leaves 4", assuming there wasn't extra space in the first part, we can fit in 28 platters for a 4.7 TB capacity.
Because the platters are smaller, the access times should be faster. It may even be possible to stick another spindle into the system (more than gaining back lost capacity), and maybe more platters in a stack for higher capacity.
Practical issues: power, heat, complexity. MTBF is probably lower. Non-failure error rates may be higher. Heat and power may be higher. The 0.75" platters may be impractical because of physical reasons. There may be some fundamental problem with having a 4-inch high 0.33" diameter spindle.

Re:Sideways

[alva_edison]

Looking at it another way. If you take a microdrive at 1.42" x 1.65" x 0.197", and shrink it to 1" x 1.23" x 0.197" you could fit ~81-84 in the same space as a 3.5" drive. The top end microdrive had an 8GB capacity. The top 3.5" drive at the time had a 750GB vs 10TB now, a 33.3x increase. The size decrease is something like 0.25"/0.67" = 0.37x. Multiplied together, we can expect the stack to have somewhere between 6.5 TB and 8 TB.

not gonna happen

[Anonymous Coward]

OK lets start thinking about what this means. So if you start making more platters and more heads that raises the cost. It also increases the chance of a head crash. More and more platters and more and more head eventually lead to a platter that doesn't even have to turn and just heads that right and read to a stationary platter.
Hence you essentially have a solid state disk. Why should they spend money trying to improve a dying technology. Solid state disks will eventually take over there too. They should be looking at different forms of solid state disks.

Waste creator?

[Anonymous Coward]

Now they are just going to make it taller and when it fails, will they just dump the whole "tower" of platters? Or are they going to fix it (expensive)? Seems like a method for maximizing waste. This would have made sense if the platters were bigger in diameter, making the pack more reliable as well.

Re:Waste creator?

[Anonymous Coward]

A big chunk of aluminum seems easier to recycle than a bunch of drive, each with their own controller and casing.

More platters = higher failure rates

[micron]

More platters yields more heads. More components to fail. This will increase failure rates for these drives at a given capacity over a similar capacity 3.5" drive with a lower platter count.

Spinning media is still hard to beat on price. Desktop 7200 RPM drives are at $.03/GB. "Enterprise" 7200 RPM SATA at volume is between $.03/GB and $0.05GB. Cheap SSD is around $0.60/GB to $1.20/GB.

A lot of data is still cold. At volume, this price difference matters a lot.

Re:More platters = higher failure rates

[Anonymous Coward]

Enterprise SATA 7200 RPM drives are closer to $0.05-$0.06. Enterprise SSDs are about $500-$800 per 1TB now, which is down to $0.50-$1.00 per GB. Consumer SSDs are about 1/2 of that price (as low as $0.25/GB).

15k Enterprise SAS drives are around $0.23/GB. Which puts enterprise SSDs within striking distance (more IOPS, less power, less noise, less heat, better density) to obliterate the 15k SAS drive market. The biggest 15k SAS drive you can buy is only 600GB, or you could put a 2TB 2.5" SSD into that slot.

The days of 10k RPM SAS drives are probably also numbered...

Turning HHDs into SSDs?

[Waccoon]

I don't understand the logic of sacrificing storage capacity for seek time. In which case, you merely end up with an incompetent SSD, an defeat the whole purpose of having a HHD in the first place.

Wouldn't it make more sense to leverage the whole advantage of a HHD and go strictly for capacity, and use more intelligent caching or more hybrid technology to reduce seek time? You can already fit a lot of platters into the 3.5" format, and stuffing more hardware into a single enclosure will probably result in too many other trade-offs to be worth it, like reduced reliability. That's especially true if the new drives are made in such small quantities that economy of scale makes the 3.5" format a better choice in the end, anyway.

Seems like another attempt to force a new format on the market, whether it's needed or not.

Bring back Univac Drum Storage & IBM 3330!

[DutchUncle]

Storage drum systems had many heads, arranged in a spiral around the drum so there was time at the end of a "ring" to select the next head. (Apparently nobody thought of making a straight line of heads and spiraling the data.) One of the later models of IBM multi-platter disk drives had 2 sets of head arms. All of these are mechanically complex, which is part of the reason for RAID (Redundant Array of Independent Disks) (redundancy of course being the other). Instead of trying to make one disk drive bigger and/or put more heads on it, make the disk SYSTEM bigger with modular increments of the then-current technology, and get more heads by spreading them across the modules.

There is a tradeoff between system complexity and speed. Highest performance has been attained on specialized systems with intelligent distribution of indices and data, particularly as differentiated by activity, across different levels of storage media; but there is benefit in maintaining a simple generic model of storage, particularly across replicated networked storage systems, so that multi-server and/or back systems need not be identical. One can envision managing active data in many gigabytes of RAM, backing it to SSD, further demoting it to HD as activity decreases; but this involves a lot of system complexity

Remember, though, that the 3330 10-platter drive held a whopping 200 MB, and a controller managing five of them could get all the way to 1 GB - when today I have a 64 GB microSD card in my camera that weighs nothing and runs at 50 MB/sec.

Ugh just what we need...

[kungfuj35u5]

a proposal for something that decreases the reliability of mechanical disks even more. I don't want higher bit rates to deal with in addition to the stream of other unpredictable failure modes associated with these things.