SATA 3.0 Release Paves the Way To 6Gb/sec Devices

An anonymous reader writes "The Serial ATA International Organization (SATA-IO) has just released the new Serial ATA Revision 3.0 specification. With the new 3.0 specification, the path has been paved to enable future devices to transfer up to 6Gb/sec as well as provide enhancements to support multimedia applications. Like other SATA specifications, the 3.0 specification is backward compatible with earlier SATA products and devices. This makes it easy for motherboard manufactures to go ahead and upgrade to the new specification without having to worry about its customers' legacy SATA devices. This should make adoption of the new specification fast, like previous adoptions of SATA 2.0 (or 3Gb/sec) technology."

Comments

isn't it time for

[markringen]

isn't it about time for us to switch to SAS? (Serial Attached SCSI)

Re:isn't it time for

[Wesley Felter]

No, because SAS will always be more expensive than SATA.

Re:isn't it time for

[Brian Gordon]

I'd say if it's bandwidth we're after, we shouldn't be reducing the number of signal lines. Do things in parallel instead of serializing everything and depending on astronomical clock speeds. Obviously PATA is obsolete but especially with the rising importance of multiprocessing we should be focusing on more parallel solutions, perhaps allowing multiple reads at a time on different lines of the connector.

Re:isn't it time for

[Anonymous Coward]

Agreed. What we need to see is some form of in-drive RAID (as a comparison only, not actual implementation), where there are multiple INDEPENDANT heads reading/writing on multiple platters all at the same time, with each head having it's own independant I/O line on the connector.

That would be cool.

Re:isn't it time for

[kaiser423]

You do realize that at either end of a Parallel link you'd have to re-serialize right? That's what PATA does. So you still need the high clock rate regardless of how much you parallelize it on the wires. That's extra hardware, and another piece the needs to be be really fast. Then you also have issues with maintaining clocking integrity over parallel lines, which gets tricky at high data rates.

Right now, our technology is better in going pure serial. In the past, it was parallel. It might swing back and forth a couple of times between the two in the future. But make no mistake: right now, on commodity hardware for drives connected via cables, serial is pulling ahead in the speed war.

Re:isn't it time for

[Wrath0fb0b]

The problem with parallel is that you can't crank up the clock speed because you have to make sure that the signal on each line is combined with the ones from the other lines that were sent at the same time. This limits how fast you can send the send the bits (if the time being bits is comparable to the skew time, the receiver will not be able to reliably reassemble the data) and how long the interconnect can be (skew being linearly amplified by length). It's not for nothing that PCI has been replaced with PCI-E, PATA with SATA, SCSI with SAS. USB and IEE1394 would be impossible with parallel. Serial communications are more reliable and more scalable (one big exception -- wireless RF, but that's not what we are discussing here).

Multiprocessing, incidentally, has nothing to do with it -- the software interface to a storage device hides all the implementation details (PATA/SATA, for instance) anyway. The hard part in multi-threading IO-intensive apps has quite a bit more to do with latency issues and atomicity guarantees (the complete lack thereof) rather than the inability of the storage device to do 2 things at once (which, for a physical disk, is impossible anyway, meaning that it would have to back-convert into a serial process anyway).

Re:isn't it time for

[hardburn]

Why? Do you have a hard drive that can even saturate a SATA I bus?

Re:isn't it time for

[Nimey]

The problem with the parallel approach is the difficulty in ensuring parallel signals get to their destinations at the same time.

Maybe in the future we'll figure out how to take today's high signaling rates and parallelize them, but the engineering choices made right now are for good reasons.

Re:isn't it time for

[stinerman]

That's right, but what the OP is (I think) saying is that we really should just look at overall performance. Generally speaking, you can get more done per clock cycle over a parallel interface than over a serial one. And, of course, what makes serial interfaces beat parallel interfaces (when they do win) is their ability to be clocked at a higher rate.

Re:isn't it time for

[ard]

Perhaps due to the fact that there is a performance difference between scsi and ata?

But, with an increased volume of SAS drive sales, the price difference should be smaller.

Re:isn't it time for

[Ilgaz]

If you are doing large database work, redundancy needed or 2K/4K video work, you may need SAS. In fact, you would still boot OS and Apps from a serial ATA device and use SAS for program (database, movie etc) data. SATA and SAS have compatible connectors for that reason. They don't really replace each other.

Of course, SAS is really expensive but for example, if you are at a professional studio which speed may actually earn you more money, you wouldn't care.

Interestingly, even SMART like features of SCSI doesn't replace actual SMART in SATA. Both have different powers. It can show how different directions IDE and SCSI went

Re:isn't it time for

[Ilgaz]

Yea, while swearing at Apple 24/7 for giving SATA1 with Quad G5 Workstation (most expensive G5), I purchased a very nice performing Western Digital Caviar 1TB drive having 32MB cache. It took a while to figure that I can't really saturate SATA1 bus, even with "fill with zeros" (format) of OS X, it went up to 140MB/sec. Of course, Apple expects me to buy a ATTO like high end card if I need more bandwidth.

What matters is SSD, that is why they release the spec right now. If you have enough money to setup a very high end (not toy-like) SSD right now, you will see SATA2 is the bottleneck. People were already talking about a different standard or even getting rid of SATA alltogether for them.

Re:isn't it time for

[morgan_greywolf]

Actually, there really isn't much difference. The main difference is that hard drive manufacturers build their SCSI/SAS drives better than their IDE/SATA drives, because most SCSI/SAS drives are going into servers.

The performance difference historically was much faster and that's the reason why SCSI is used in server hardware, but now it's mostly a matter of economics and pricing.

Re:isn't it time for

[quenda]

at either end of a Parallel link you'd have to re-serialize right?

Why? At the disk end, enough platters/heads will give you bits in parallel. Just buffer
each to allow for skew, and you can read a byte on each clock tick.

And why would you need to re-serialise at the main bus end?

NOT that I am saying parallel is a good idea.

Re:isn't it time for

[vadim_t]

It's been tried, and didn't work well.

The drive heads are some of the most expensive parts of a hard disk, so it raises the price considerably. Then you get higher power usage, heat generation, decreased reliability, and higher complexity in exchange for the extra performance.

The problem is that normal people don't look at speed, they look at capacity. So they won't buy the expensive drives. And the people who do look at things like bandwidth and latency are already running a RAID and benefitting from multiple heads already. They're also unlikely to want something that's less reliable.

Re:isn't it time for

[hairyfeet]

What i think he was trying to get at is this: say you have a 4 platter HDD. Instead of having the heads all working as one you feed each of the platters off a different SATA line. In essence you would have a single RAID drive doing RAID-0. Which might just work, but it would depend on how much the controllers and extra motors would add to the cost.

What I want to know is what happened to the hybrid idea? I though that was frankly the way to go. Much in the same vein as the WD "green" drives you could have the things that are less likely to change(Say the core OS) stored on a nice fast 16Gb of SSD, while having pictures, videos and all the other programs and files stored on a much larger but slower HDD. I doubt it would even be hard to set up. Just have the drive come with an installer that after OS install uses Junction points to point Documents and Settings, Program Files, and all other non core OS files to the HDD. For all other Operating Systems(or if you don't want to use the installer and instead want to DIY) it would simply look like 2 drives-the smaller SSD OS drive, and the larger HDD storage drive. This would allow everyone to take advantage of the speed of SSD while not having to sacrifice storage space as they do currently.

There also needs to be a reliable optical or tape based storage device marketed to the masses. With DVD and even BD(which is still way too expensive for the masses and has too many DRM problems) there simply isn't enough storage to backup the mounds of HDD space that even the cheap machines are coming with nowadays. External HDDs should be only viewed as a stopgap because you still have the inherent problem of dealing with a mechanical medium. There needs to be a simple way to back up large amounts of data and simply "stuff it away" until needed such as what servers have with tape. But we really haven't seen any medium targeting the home user since DVD, and with the mounds of data being created by cameras, camcorders, etc it really is long overdo IMHO. Because all the speed in the world doesn't help if you can lose it all simply by accidentally dropping the external HDD.

Re:isn't it time for

[Theovon]

And what you clearly missed from the post you're responding to is that the clock rates that you can get from serial are so much higher than what you can do with parallel that it more than offsets the disadvantage of serialization.

There are two things that limit the speed of parallel interfaces. As the GP mentioned, one is signal skew. The clock rate has to be slow enough so that the receiver can sample all data lines at the same time and get them all within the data eye. The second is that the data lines are single-ended, meaning that there's only one wire per signal. The clock rate has to be slowed down to ensure that the signals have reached full on or full off at the other end and that they're noise free.

High-speed serial interfaces use DIFFERENTIAL SIGNALLING. The signal is transmitted over two wires that switch in antiphase. You decode them by comparing them. This has several beneficial effects. One is that noise affects them the same, so even if they're both offset by noise, they compare the same. The other is that now you don't have to wait as much on the effects of resistance, capacitance, and inductance. You can reliably decode the signal before the transitions are complete. (Look up "slew rate".)

So, using the same basic silicon technology, you can get a single differential pair to transmit data MUCH faster (in bytes/sec) than you can with parallel. It's interesting to see how technology transitioned from serial to parallel (wider means more bits per second), back to serial. The reason they didn't just stick with serial was that they just didn't have the technology to make the I/O drivers go that fast until recently.

IIRC, A 1x PCI Express channel is a single differential pair for data. (I think there's a side band channel and some other stuff.) This is just like DVI and SATA. 16x PCI Express is sixteen 1x channels. The trick here is that although data is interleaved across all 16 channels, those channels are not syncronized with each other. They are out of phase, and the the data is just put back into phase at the receiver.

Re:isn't it time for

[Barny]

Firstly, multi platter is not the best, increased heat, increased complexity all increases rate of failure.

Secondly, you now are making a standard for the number of platters/heads in a drive, in reality everyone wants something different (reliability over density).

Re:isn't it time for

[fuzzyfuzzyfungus]

Well, it is increasingly irrelevant, with the gradual rise of SSDs as the tool of choice for very high IOP workloads; but 15k RPM drives have long been the exclusive domain of SCSI/SAS. There are a handful of 10K RPM SATA drives; but everything else is lower spindle speed.

For drives of equivalent spec, on SAS, on SATA, same spindle speed, I suspect that it is largely marketing fluff and a few firmware tweaks; but 15k RPM vs. slower is a nontrivial difference.

Re:isn't it time for

[bofh29a]

There are cheap SAS drives and expensive SATA drives, eg

1) Seagate 1TB SAS, 7200rpm -USD217, that's USD217/TB (http://www.provantage.com/seagate-st31000640ss~4SEGB03W.htm)
2) WD Raptor 300GB SATA, 10krpm-USD240, that's USD800/TB (http://www.provantage.com/western-digital-wd3000glfsrtl~7WNDG2LT.htm)

Hence, it's not merely the interface that determines the price.

Re:isn't it time for

[gmuslera]

What about one of this ones? They could probably take good advantage of a 6 Gb/sec speed

Re:isn't it time for

[Penguin Follower]

For drives of equivalent spec, on SAS, on SATA, same spindle speed, I suspect that it is largely marketing fluff and a few firmware tweaks; but 15k RPM vs. slower is a nontrivial difference.

I agree completely. We've got two SANs at work... the older one is full of U320 10k RPM drives and the new SAN is all 15k RPM SAS drives. The new SAN leaves the old one in the dust (and has 20TB more space, too! :D).

Re:isn't it time for

[AHuxley]

Who dares windows.
from Qui audet adipiscitur

Re:isn't it time for

[Antique Geekmeister]

And the OP is, frankly, unaware of the history of SCSI and PATA. Those big wide cables are deprecated for many reasons: one is their expense, another is their fragility, and another is the incredible variety of vaguely distinct, and often stupidly different, specifications for such broad interfaces. I had to deal with that debris, for decades, and it was extremely painful.

The amount of time saved in consistent, small interfaces having fewer things to screw up is enough, by itself, to make up for the expense of any drives lost from the fragility of the SATA connector. I remember the amazing crap shoot it used to be to design a SCSI chain of devices, the awful incompatibility and expense of the cables even for what were nominally the same type of SCSI, and tendency of those connectors to bend pins or fail under stress.

Give me SATA (and its low cost peer for external devices, USB), any day over the technically superior but less consistent SCSI and firewire.

Re:isn't it time for

[Guspaz]

Yes. I do. My single drive has an average sustained transfer rate of 230MB/s. A SATA1 bus would severely constrain the performance of my drive (an Intel x25-m).

There are numerous other SSDs on the market whose manufacturers focused on higher sustained performance rather than random access performance that already hit the 300MB/s wall of SATA2. And I expect that Intel's next series of drives will do the same. SATA2 is woefully unprepared for the very near future, let alone the present; it's slow enough to already be constraining high-end performance.

Re:isn't it time for

[rcw-home]

The second is that the data lines are single-ended, meaning that there's only one wire per signal.

Well, this may not be exactly what you were getting at, but I'd like to split hairs here anyway, and divide this into two separate issues that SATA/SAS resolved.

For best results it's important to model the cable as an RF transmission line, with a specific impedence. An ideal transmission line has the important qualities that all the energy you send from one end will arrive at the other, and none will be reflected back to you. To get reasonably close to this ideal, we space the wires we use a fixed distance apart (in relation to the wire's diameter), choose our dielectric (insulating material) carefully, use terminating resistors at both ends, and keep the line a simple line (no tees, etc.)

For those of you who cut your teeth on parallel SCSI, 10base2/10base5 Ethernet, or Apple LocalTalk, you'll wax nostalgically at just how much of a pain in the ass this was.

For those of you who have only messed with parallel IDE, you'll wonder why you never had to deal with this. The reason is that IDE cheated a little bit - they only terminated the controller (motherboard) side of the bus, and let the signals reflect off the other end. This left only a master/slave/cable-select jumper to infuriate you - but it also limited how long an IDE cable could be and prevented them from jacking up the clock rates on it.

SATA/SAS fixes this for good by limiting you to one device per cable ("port", not "bus"). Both ends are hard-wired to always terminate and any cable problems are limited to a single drive.

The other issue you may have been referring to is balanced (differential) vs. unbalanced signalling (where one wire is held to ground and the voltage read off the other wire). Electrical engineers do commonly call unbalanced signalling one wire because ground is so boring that they never bother to connect it on their schematics, but it does have to be connected in real life and coax Ethernet/most old SCSI/Parallel IDE/RS-232/VGA still used two wires per signal. Balanced/differential signalling (LVD/HVD SCSI, SAS, SATA, 10/100/1000baseT, USB, telephone lines, T1 lines, LocalTalk, etc.) allows for the can't-imagine-life-without-it common-mode noise rejection technique you describe.

Re:isn't it time for

[chef_raekwon]

i'd have to say, however, that comparing two SANs like you've done, doesnt necessarily mean that 15k is better. i would presume that if your SAN has 20TB more space, it likely has more disk. If it has more disk space, it has likely been partitioned out for more IOPS, by "virtue" of having more disks. More disks = more IOPS, especially in a RAID 1+ 0 configuration.

Not that I'm disagreeing with you -- 15k drives are definitely faster, its just the logic. I guarantee I could build a faster SAN with 10k rpm disks (with more disks), than with a nominal amount of large sized 15k disks.

Re:isn't it time for

[guruevi]

Apparently you are not completely up to snuff with your jargon there.

I have worked with the guts of computers long enough to have known ESDI drives (in the PS/2 no less) those had as far as I remember serial data lines (and a separate control line to control head movements). Then came SCSI and IDE (later standardized as ATA, faster versions as EIDE or ATA-2, for CD/DVD/ZIP drives ATAPI and recently known as PATA) which were parallel versions.

The first SCSI drives I used had 8 data lines (SCSI-2) - you could even make your own cables for those things, very robust. Later SCSI's (Wide SCSI) had 32 data lines and a very wide connector with thick cables that would have the whole side of your case covered with cables if you were putting in a dual controller setup - sometimes those cables would have so much tension and take up so much space they wouldn't stay in the drive and then you could start rerouting the whole cable again to find a 'better' way.

ATA had less of an issue as the cable wasn't as wide nor thick but you could only get 2 devices on a cable and the one designated slave (usually CDROM) could tie up your bus and use valuable time so you should've ran a separate cable for each device.

The problem with both Parallel SCSI and Parallel ATA was that you could only drive it up to a certain speed before you would get synchronization issues between the data lines. Serial is much cheaper for that as you can drive up the frequency without caring too much about the synchronization.

Firewire and USB have always been serial. Firewire is technically superior and also more expensive than USB (same as SCSI was far, far, far more superior than any IDE installation for the same reasons) because the devices (both host and target) require internal controllers so as to not tie up the CPU. SCSI also required (sometimes manual) terminations and (before SCSI-3) SCSI ID's

SAS is Serial Attached SCSI, just like SATA is Serial ATA. SCSI (this time serial) is again, more superior but also more expensive than ATA and allows much more flexibility (you can for example attach SATA devices to a SAS controller, not vice versa) and SAS can maintain multiple drives on a single cable while SATA is limited to one device per cable.

Give me Firewire and SCSI over USB and ATA anytime.

Re:isn't it time for

[jhol13]

Yes http://www.sun.com/storage/flash/module.jsp.
It does not use SATA, but something JEDEC will(?) standardise.

I think it is very interesting idea, whether it will take off, I do not know.

Re:isn't it time for

[rivaldufus]

maybe the cache could...

Re:isn't it time for

[Anonymous Coward]

IIRC, A 1x PCI Express channel is a single differential pair for data. (I think there's a side band channel and some other stuff.) This is just like DVI and SATA.

Actually, I remember looking this up just last night. Each PCIe lane consists of two transmit pairs and two receive pairs, for a total of four pairs. DVI also uses multiple pairs, both in single-link and dual-link mode. I'm not sure about SATA.

At the electrical level, each lane consists of two unidirectional LVDS or PCML pairs at 2.525 Gbit/s. Transmit and receive are separate differential pairs, for a total of 4 data wires per lane.

Thus sayeth the Wiki.

I think the only other thing you missed was a discussion of crosstalk. IIRC, crosstalk was one of the major limitations of the old IDE cabling standard, since transmitting many high speed signals in parallel was a recipe for interference. The cables were therefore required to be of a particular length (18 inches), shape (flat), and electrical characteristics. SATA and especially eSATA aren't anywhere near so picky.

Re:isn't it time for

[upside]

Absolutely. I'd also wager the new SAN controller has more battery-backed cache than the older one. This makes a HUGE difference.

Re:isn't it time for

[nausicaa]

Well, no, not a SINGLE disk.. But, hey, I'm using a backplane/port-multiplier combo that allows me to connect 5 drives to a single SATA-connector..
(I think someone actually mentioned something like this, far above in the earlier comments)

Besides, having interfaces be ahead of the drives, performance-wise, is not a bad thing, it's actually a very good idea, so that drives can advance without hitting the roof..

Re:isn't it time for

[petermgreen]

The thing is signals running in paralell with shared timing doesn't work very well beyond a certain clock speed*distance product.

Signals running in paralell with independent timing seems at the moment to be the way to go for really high bandwidth (this is what PCIe uses and afaict SAS has the capability to use it as well) but it gets pretty complex.

Re:isn't it time for

[petermgreen]

High-speed serial interfaces use DIFFERENTIAL SIGNALLING.
So do some high speed paralell interfaces (such as some SCSI variants)

The trick here is that although data is interleaved across all 16 channels, those channels are not syncronized with each other. They are out of phase, and the the data is just put back into phase at the receiver.
Indeed this seems to be the method that gives the highest performance because it combines high clock rates with parallism. The downside of course is the complexity of buffering it into sync and then de-interleaving it.

Re:isn't it time for

[petermgreen]

SAS can maintain multiple drives on a single cable while SATA is limited to one device per cable.
My understanding was that with both SATA and SAS cables were point to point and both could support multiple drives per port with additional hardware. However SAS expanders do allow for far more drives on a port and a much more complex topology than SATA port multipliers.

Re:isn't it time for

[Antique Geekmeister]

Consistency and price beat quality by a landslide, when the quality difference isn't large enough. While firewire and USB are both serial, their example is illuminating. The less expensive, somewhat less capable USB has defeated firewire hands down in available devices and their cost. Similarly, while SAS is more capable than the current SATA, especially with hard drive transfer rates and the RPM's of hard drives to match that capability, SATA is far more widely deployed and the drives are _huge_.

For ease of installation and management, give me SATA and PATA over SCSI anyday. Stringing those drives together was a nightmare of termination, of setting device ID's, and of buying 3 external devices, all "SCSI 2", and discovering that they had 3 different types of connectors and I had to buy enough expensive cables or adapters, and terminators, to deal with all of them to be able to swap around equipment. And my goodness, the sad joke that is "auto-termination" was its own nightmare.

Give me the far less expensive and far more stable SATA internally and USB externally for hard drives, anyday, over SCSI and firewire.

Re:isn't it time for

[ta bu shi da yu]

I sure as heck hope that if you go with that option you a. use cache vaulting, and/or b. don't use the feature on disks that hold database transaction logs.

Re:isn't it time for

[hairyfeet]

How so? While RAID-0 doesn't give you any protection like RAID-5, pretty much everyone isw in agreement that it makes for some serious performance boosting. That is why some of my gamer customers(against my advice) are running RAID-0 on their gamer rigs. They do get some serious FPS on their shooters that way.

Oh, and the reason why I do not recommend RAID for the home user is most home marketed RAID controllers are shitty. I have seen the RAID controllers bone the data way too many times, even on the "hardcore gamer" boards, and ATM the quality server class boards are still too high. That is why i tell my customers just to run multiple drives, one for the OS and a second faster one, a raptor if they can afford it, for the games. Along with a buttload of RAM, of course.

Re:isn't it time for

[lpq]

No, because SAS will always be more expensive than SATA.

Why? If you break products down into consumer and enterprise level drives -- as they do within SATA drives, then presumably, you can also have consumer and enterprise level SAS drives.

So why would a SAS drive at same quality (if you compare same speed and capacity) be more expensive than a SATA drive?

It appears from this jaundiced eye, that the distinction is purely arbitrary to support significantly higher mark-ups and margins for SAS products. With SATA Enterprise products, the mark-ups are noticeably lower as the Enterprise makers have to provide real performance and/or quality beyond the 5-year warrantied consumer level products.

Measuring quality for quality, why would SAS be any more expensive than SATA? (once you've separated out any difference for Enterprise vs. consumer level quality).

Seems like price of SAS drives and parts would fall due to increased volume.

Re:Theoretical != Real World speeds

[Pinky's Brain]

SSDs are pulling a whole lot more than that ... at least when they are new ;)

6 Gb/sec? Meh

[Totenglocke]

Let me know when we hit 1.21 GW -- then I'll be excited!

What is the point?

[supervillain]

No current hard disk or even SSD can do 3Gb/sec so what is the point?

Re:What is the point?

[Jason Pollock]

Devices which aggregate themselves as a striped array behind a single eSATA/SATA interface. While the individual device may not be able to pump out enough data, they can in aggregate.

Re:What is the point?

[Lewis Daggart]

Whats the point of designing a hard disk or SSD that works faster if SATA is stuck at 3Gb/s?

Re:What is the point?

[BeardedChimp]

Exactly, it's not like technology advances or anything.

Re:What is the point?

[wjh31]

3Gb/s = 375MB/s a typical HDD should be easily capable of a third of that. So high end SSD's can probably beat 375MB/s, if not now then soon

Re:What is the point?

[Wesley Felter]

Current SSDs are very close to the SATA 2.0 limit and the performance of flash is about to double thanks to ONFI 2.0, so we can expect SSDs to quickly adopt SATA 3.0.

Re:What is the point?

[ichigo 2.0]

Actually the limit is 300MB/s which some of the new drives are very close to reaching. One more generation of SSDs and they'll be bottlenecked by SATA 2.0.

Re:What is the point?

[geekoid]

Not true. SSDs are approaching that now.

HP has an enterprise SSD that is 800MB/s (Note the large B as opposed to b). So this drive could saturate SATA 3's 6 Gb/s

Re:What is the point?

[wbattestilli]

Actually, the OCZ Vertex http://www.ocztechnology.com/products/flash_drives/ocz_vertex_series_sata_ii_2_5-ssd/ can sustain about 230MB/s. This is equal to about 2.3 Gb/sec. Allowing for the rate at which SSD technology seems to be changing, I'd say that this standard is just in the nick of time. Chances are that the next thing in high-end, consumer SSDs will saturate a SATA link. If this standard doesn't get pushed out soon, drive manufacturers will be doing ugly, proprietary, OS specific hacks to support multiple SATA links to a single device. In addition, lots of people are packaging multiple physical drives into a single SSD with an internal RAID-0 controller. These are definitely being(or soon going to be) held back by the 3BG/s SATA link.

Re:What is the point?

[Jah-Wren Ryel]

No current hard disk or even SSD can do 3Gb/sec so what is the point?

Oh yeah?

Re:What is the point?

[Shikaku]

Faster, Faster!

More, More!

I think it's some kind of nerdgasm with hard drive space and speed or something.

Re:What is the point?

[morgan_greywolf]

Faster, Faster!

More, More!

OMG, I think I just blew my wad!

Re:What is the point?

[BikeHelmet]

RAM drives would hit the SATA 3.0 cap, and they were available in 2005. The Gigabyte I-RAM could manage multiple GB of transfer per second(like all RAM can), but was capped by using a single 150MB/sec SATA1 port. :/

Luckily RAM is so cheap now, that if you really want a RAM drive, you just buy 16GB and create a software RAM drive.

Oh - but in 2005 games didn't average 8GB each. :P

Re:What is the point?

[Splab]

No, it's about not having to wait for something. I want 100mbit/s, not because I need massive files transferred around the clock, but I don't want to wait for hours when I want an update to something.

Same goes for hard drives, faster is better. Most of the time I'm doing some low intensity stuff, but just once in a while I need to move a massive file around, extract data etc. and again, I don't want to sit around for hours waiting for it to complete.

Having 300MB/sec means I can shuffle a DVD around in 15 seconds, now that could come in handy.

Re:What is the point?

[petermgreen]

Luckily RAM is so cheap now, that if you really want a RAM drive, you just buy 16GB and create a software RAM drive.
Ram may be cheap but server grade motherboards and CPUs aren't. And even when you get the server grade stuff there is still a hard limit on how much you can put in (and usually the largest size of module has a FAR higher cost per gigabyte than the one half the size)

HDDs, SSDs and ram drives with SATA interfaces and one can easilly connect 50 or so drives to a machine with multiport cards and port multipliers.

Also software ram drives are usually volatile while hardware ones are often battery backed.

Re:What is the point?

[kiehlster]

True, but the real point is to confuse marketing lingo so people can sell you a SATA 3Gbps drive as a SATA 3.0 drive and cause consumers to become even more stupid and thus eliminate the build-your-own-computer market.

Re:What is the point?

[ichigo 2.0]

While I agree that too much space is not very useful, I can't say the same about speed. Many games take tens of seconds to load, so anything that reduces loading times is a Good Thing. :)

Re:Theoretical != Real World speeds

[0100010001010011]

Gb!=GB. Divide by 8.

And you should beck your drive settings. My old IDE drives beat 20MB/s. I just checked my newest SATA drive and I got 113MB/sec in hdparm.

Ah!

[vancondo]

My bank account will be delighted that there's a reason for me to hold off buying a new system.

..that is until it see's me buying overpriced bleeding edge buggy gear again.

-
-- VCI

Re:Ah!

[prjt]

My bank account is dead.

Re:Ah!

[Ilgaz]

You don't need to buy new system. Well, even if you need SATA3 bandwidth, regular companies will release interface cards which will be better performing than "coming in mainboard" ones. I'd prefer a cache having, dedicated and configurable SATA card instead of that dumb chip on mainboard anytime.

Of course if you talk about laptop, it is a different matter.

Re:Ah!

[default luser]

Yup, SATA 3.0 plus USB 3 are both going to be released in-bulk soon. I'd definitely buy a system with these blazing new I/O options.

Re:Theoretical != Real World speeds

[evanbd]

Wow, both your numbers are wrong. SATA 2.0 has a theoretical transfer rate of 3Gb/s, not 3GB/s. It also uses an 8b/10b encoding, so 3.0Gb/s translates to 300MB/s. Data throughput will be less than that, thanks to control protocol overhead, though the overhead is very small.

Modern drives do seriously better than 25MB/s. Seriously, go look at benchmarks. Also, SSDs, which are a very real design influence on things like SATA, are already getting close to the 300MB/s mark.

Re:Theoretical != Real World speeds

[markringen]

ssd's will probably end up being connected to a form of ram socket with an on-cpu controller (like system ram) in the future. eventually flash can be half as fast as system ram, so there is no real reason not to have it connected to the CPU.

Worth noting

[earnest murderer]

The spec as we have seen with most other transfer specs have little to do with real world device designs. Hardware interfaces (much less devices) languish in the "has to cost less than x per part" hell... But you bet your ass they'll put a SATA 3.0 up to 6GB per second label even though the actual device isn't designed to transfer more than a fifth (peak) of the spec. data rate.

Re:SSD

[earnest murderer]

Probably not. Even if you had a device that would supply the (most? if not all) commercial interfaces aren't actually capable of moving it that fast.

Re:Theoretical != Real World speeds

[Alystair]

Just saw a 4x SSD Raid-0 setup pull 350MB/s, so I'd say this is a good direction :-)

This is NOT SATA 3.0, children! Smarten up.

[Anonymous Coward]

http://www.serialata.org/developers/naming_guidelines.asp

Here's a clue: If you have to post a web page explaining the proper way to refer to your products, your products are poorly named.

Here's another clue: If there's a shorter/easier/faster way to refer to your product, people are going to go with that. Insisting that they do otherwise indicates delusions of grandeur.

Get the hell over it already.

I hope they make the plug stronger

[pembo13]

I've lost 3 drives due to plugs breaking off into the SATA ports on the 3.5" drives

Re:I hope they make the plug stronger

[ColdWetDog]

I've lost 3 drives due to plugs breaking off into the SATA ports on the 3.5" drives

Agreed, that's the dumbest physical connector I've seen in the longest time. I'd like to take those broken bits and shove them up the fingernails of the engineer that designed it.

Re:I hope they make the plug stronger

[Anonymous Coward]

I've lost 3 drives due to plugs breaking off into the SATA ports on the 3.5" drives

Agreed, that's the dumbest physical connector I've seen in the longest time. I'd like to take those broken bits and shove them up the fingernails of the engineer that designed it.

Obviously, you have never used an HDMI connector.

Re:I hope they make the plug stronger

[zonky]

I raise you SCART. (Thankfully now disappearing.)

Re:I hope they make the plug stronger

[grommit]

Maybe you should stop using a hammer when plugging in a new hard drive?

Re:I hope they make the plug stronger

[yachius]

Same here. I plugged in a drive a few weeks ago with a regular straight cable and bent the cable up to fit in the case and the connector promptly snapped off.

Re:I hope they make the plug stronger

[the_other_chewey]

I've lost 3 drives due to plugs breaking off into the SATA ports on the 3.5" drives

Agreed, that's the dumbest physical connector I've seen in the longest time.

Nah - that would be SCART
Most awful connector ever, especially in comibation with poor manufacturing.

Re:I hope they make the plug stronger

[rivaldufus]

That, and the super glue on the SATA cable connector isn't helping much - especially when trying to remove the cable.

Re:I hope they make the plug stronger

[drizek]

Connectors should be made so that if something breaks, it should be the cable, not the device it is connected to!

Re:I hope they make the plug stronger

[ParanoiaBOTS]

Maybe you should stop using a hammer when plugging in a new hard drive?

They should probably also remove the anvil they are hanging off it once it is plugged in...Seriously how can you snap them? If you have that much tension in the cable you are doing something wrong.

Re:I hope they make the plug stronger

[demo]

FFS. Breaking the connector when you're putting forces on it outside its rated spec is not a problem with the connector, it's a problem with the idiot using the connector.

Well, no, when several people have the exact same problem with something, it's usually a problem with the design.

Re:I hope they make the plug stronger

[thexile]

I concur as well. I have a drive in RAID 0 that has a broken connector (the plastic broken however the metallic connectors are still attached). Thankfully, it's still working.

Re:I hope they make the plug stronger

[petermgreen]

Whats so bad about scart? it's big enough to be easy to hand solder, Reasonablly good at staying in it's socket, doesn't tend to break if it gets yanked out.

It's a bit bulky but thats only really a big deal with smaller/thinner flatpanels, not with the CRTs that were the norm when scart was introduced.

Re:I hope they make the plug stronger

[petermgreen]

PC cases tend to be pretty cramped environments and since sata have a mixture of connections with very different force requirements. This a recipie for stuff getting broken if you aren't VERY carefull. It's especially bad in mixed SATA/PATA systems.

Re:I hope they make the plug stronger

[fruey]

Only improved via HDMI and other digital interfaces... SCART is the only standard RGB and/or YC and/or composite with stereo audio and a couple of spare pins. It's pretty much still standard even on the latest greatest flat panels too.

Until HDMI sorts out issues with possible DRM and other compatibility issues, RGB analog with one SCART is way easier than five RCA/cinch plugs...

Re:Theoretical != Real World speeds

[Alystair]

Eh sorry, that small number is from IOMeter.. CrystalMark reports 1504MB/s for sequential reading. 75MB/s for 4k random writes ain't too shabby either :)

Re:Only one problem with this:

[mikael]

Maybe they will double the disk memory cache to something larger like 16MB or 32MB. Even with only 8MB it really amazes me that just the disk drive has more RAM memory than PC's from a decade ago.

Re:Only one problem with this:

[vadim_t]

For hard disks, making them much faster isn't really possible. The disk needs to spin faster, or the information needs to be packed more tightly. Currently advances are mostly in the packing, but aren't reaching yet even SATA II levels.

Hard disks will get a slight benefit though because they have a cache and they can transfer data from or to it faster than the platter can handle.

For SSDs, even exceeding SATA 3 is perfectly possible by simply internally parallelizing requests. Also, for SSDs, the interface's latency is probably a fairly significant part of the time it takes to service a request.

Re:Only one problem with this:

[Anonymous Coward]

how about they put some equivalent effort into speeding up the actual output of devices that use the interface?

SSDs use the interface, and they're getting close to hitting the 300MBps throughput mark (maximum after sata overhead).

There are also several external raid enclosures that use eSATA and appear as a single high-throughput drive to the onboard sata controller.

Sata Smata

[nurb432]

What about us using MFM drives with removable platters?

Re:Sata Smata

[hurfy]

We are still looking for serial cables with 8ga wire......

Re:Sata Smata

[morgan_greywolf]

We'll get off your lawn now.

Re:Sata Smata

[Thing 1]

LOL, thanks. (And stay off!)

Stupid

[TheParadox2]

I think in a years time frame, we could see the 6 Gb/s passed with the way SSDs are going. To make this standard is dumb. If we're looking for speed, SATA 6Gb/s is not it and this ancient CHS scheme has to go to accommodate a better way to map, access and control data. Ultimately, we need to have these devices understand & control the file system. (Trim does this for SSDs) For example: The OCZ vertex nearly saturates the 3Gb/s mark already. They only way the drives 'fail' to accomplish this sustaining speed is with random writes, typically which occur when writing data to a spot marked as available when the NAND isn't zeroed, it either has to re-zero or move on. If the drive knows that the OS is deleting a file (not marking the site, as available) then the drive can zero automatically without you noticing. Its only in certain conditions, these drive don't Consistently perform at peak performance: Free space not consolidated, Free space not zeroed, Swap file creates random writing (slows performance), Indexing is now useless with .1 ms seek times. Using write filters, or something that converts random writes to sequential writes (through buffers, caches or drivers) greatly enhances speed, such as the MFT Software or even windows SteadyState for the devices. I like the idea of the 'RAM socket' interface as someone stated above. These devices i think work better in a parallel manner. Most work like this internally anyway.

Re:Stupid

[afidel]

I think the most likely outcome is SSD's move to something like ExpressCard, a physical spec which extends the PCIe bus out to the storage. The drives will show up as a SCSI/SATA controller AND a virtual disk attached to that controller so that the software layer doesn't have to be changed.

Re:Stupid

[symbolset]

The attach of the next generation of internal storage media is PCIe. And the attach of next generation expansion storage media is... external PCIe. 2-3 years out flash storage will be cheaper per TB than spinning disc. Sometime before that most folks will realize that millisecond latency is not as good as microsecond latency. So... Yeah, I agree with you.

Re:Stupid

[maxume]

Care to flesh out your prediction? I'm looking for info, not sniping (last time I looked at the trends, it wasn't going to be anywhere near that soon; I didn't look real close, and I guess if spinning disc hits a wall it would speed things up).

Re:Stupid

[symbolset]

For PCIe flash density is doubling and the price is quartering every nine months. With current price for capacity at less than $2500/TB, you can quarter that twice and get to $150, and double it twice to get 4TB. So... at the current rate of change it's inevitable. For density you will soon buy a 1TB flash drive in a 2.5" form factor. I could put four or six of those in the space of a full length add-in card and thermals are not problem so we're nowhere near the limit.

I don't expect the trend to continue to be this steep, but 3 years is a long time. If it double/quartered half as fast, that's time enough to get 'er done. In the enterprise market the performance and power savings are already there, and the storage density is almost there, and the money isn't as important - those poor folk pay thousands of dollars for one 450 GB FC drive.

The reliability and durability figures need to come up, but that's well underway also with most vendors taking an internal adaptive RAID approach that puts more of the burden on the board logic and less on the host - holding some storage in reserve to wear level the last block.

The final pieces are iSCSI, 10Gb Ethernet and SAN software that converts direct attach storage to SAN with all the features we know and love - because those SAN head boxes are just general purpose computers with fancy software and lots of direct attached storage. These are all here now. Spinning disk might become the next version of backup tape for a little while, but even for that it's doubtful as they're pushing the quality limits pretty hard to get the density they're getting now and people are getting tired of drives that fail so much.

So yeah, 2-3 years I think. For some people, sooner. The whole world is changing so fast it boggles the mind. You can already buy a workstation that supports 16 threads, and next year we get Larrabee.

Re:6 Gb/sec? Meh

[Zerth]

Surely you mean 1.21JW

1.21 Joule Watts?

WTF is 1.21 m^4*kg^2/s^5 good for?

Re:Theoretical != Real World speeds

[hardburn]

Or just be RAM with a battery backup.

Re:6 Gb/sec? Meh

[DimmO]

Time travel?

Re:SSD

[Wrath0fb0b]

If my understanding of the technology is correct, the seek time on most hard drives already limits drive access speed to typically be slower than 3Gb/sec. Would this rely on a transition to Solid State Drives for any noticeable difference in performance?

The seek time has nothing to do with the throughput. The seek time refers to the latency between when a read command is issued and when it begins to be fulfilled. The throughput refers to the data transferred per unit time during fulfillment.

Here's a nice car analogy for those of us in New England -- consider the Mass Pike versus I-93. The Mass Pike has a very long seek time from the onramp because of the toll lanes (and the mouth breathers that won't get a transponder even though they are now free and clog the automatic lanes) but once you get on the highway, you can go 80 MPH until your exit. On I-93, by contrast, you can get right on, but you will be going 30 MPH for the duration. Of course, if you drive down to CT and get on I-84, you have a low-latency AND high throughput highway but if you drive too far down to, say, the Bronx, it becomes high-latency and low throughput.

Forget Heads...

[RudeIota]

where there are multiple INDEPENDANT heads reading/writing on multiple platters all at the same time

The entire idea of 'heads' should be forgotten. Mechanical drives should be sent to oblivion and we should welcome your idea of parallelism on solid state solutions.

Re:Forget Heads...

[Clay Pigeon -TPF-VS-]

Did I miss the memo that says flash no longer has a limit on how many times it can be written upon?

Re:Forget Heads...

[camperdave]

While all flash is solid state, not all solid state is flash.

Re:Forget Heads...

[vadim_t]

It has a limit, but it's hard to reach even with cheap flash, and almost impossible with good flash unless you're doing something really unusual. And the bigger the disk gets the harder it is to reach it, due to wear levelling.

Also, hard disks are not eternal by any measure, and will fail mechanically, and often without warning and in a much less predictable fashion.

Re:Forget Heads...

[Barny]

Hard to reach? Try installing vista on one, and letting it use it for temp files, system logs and page file.

In fact, thats all on by default! Not exactly hard.

A better idea is to install on a normal HDD, then copy all the files from program files into a SSD and mount it in an empty folder and point windows to that for its program files directory, not the neatest solution, but stops the drive dieing too young and gives a marked speed boost.

Re:Forget Heads...

[vadim_t]

If you're swapping enough to kill a SSD, you're doing it very wrong. Get more RAM, it's cheap.

SSDs do wear levelling, so even if you have a limit of 10K erases that's a per-block limit, and there is a LOT of them on a modern SSD. To kill it, you'll need to make a quite serious effort.

For instance, suppose you leave only 16GB free on your SSD, which has a 128K erase blocks. That means there are 131072 blocks the SSD can recycle, not including whatever amount it reserved internally. To kill those, you'll need to make 1310720000 block erases. That's a pretty big number.

Re:Forget Heads...

[Barny]

Lets see, with firefox, steam and the performance monitor open, I am clocking about 4MB/min of writes (thats with not a lot of firefox windows open too, its using about half of that amount).

Note since most windows use 64K clusters, every time 64K of data is written, a 128K block needs to be erased and written.

Oh, and my work PC (the one the measurements were taken on) has 4GB ram under vista 64, my home is double that.

Re:Forget Heads...

[vadim_t]

Ok, so let's say those 4MB/minute (IO writes/s would be a better measure) are made from 64K requests. So that's 64 requests/minute, or about one a second.

That's not terribly high, so let's double it to 2 requests a second.

1310720000 max block erases, at 2 per second will last 7585 days, or 20 years.

This assuming a MLC drive with 16GB available for reallocation for it. If you use a SLC drive, you probably won't live long enough to see the disk wear out, and even with MLC it's doubtful you're going to keep the same drive around for 20 years. I think that 20 years ago you'd be running a 386 or a 486, and have maybe 200MB of disk space, and can't even plug in a hard disk from back then into most modern computers.

Re:Forget Heads...

[AHuxley]

6.40 GB/sec should be enough for anybody.

Re:Forget Heads...

[Barny]

You are very very kind to windows, that 4MB/min of IO was across about 20 different processes, most of which were writing a few bytes a second, not nice neat 64K writes (or even, as you add double, 32K writes).

Re:Forget Heads...

[Barny]

Btw, a good read on what some of the bad things could be for them.

http://forums.slizone.com/index.php?showtopic=29361

Re:Forget Heads...

[rcw-home]

Did I miss the memo that says flash no longer has a limit on how many times it can be written upon?

Apparently. At the very least you are making "perfect" the enemy of "damn good".

Re:Forget Heads...

[w0mprat]

A better idea is to install on a normal HDD, then copy all the files from program files into a SSD and mount it in an empty folder and point windows to that for its program files directory, not the neatest solution, but stops the drive dieing too young and gives a marked speed boost.

YABI (Yet A Better Idea) is to use the oh-so-sensible Linux strategy of separate partitions for data type.

Put your windows and program files folders on the SSD, and link the \users\ folder to a seperate HDD NTFS parition. Locate your page file on the first NTFS partition of a nice fast HDD. Have it as a fixed size so it does not become fragmented.

This has the advantage of temporary files under user profiles being written to a HDD. This is not necessarily slower as the extra drive is *additional* bandwidth to your system accessed in parallel.

Another performance method that goes hand in hand with this method is short stroking, basicly have your system in the first 10% or 100gb of a hard drive - whichever is smaller and ensure only seldom accessed media files etc are using up the space. This stops your operating system rotting over time with scattered file placement. This can boost your access times massively.

My strategy is SATA0 - 32GB OCZ SSD - Windows 7 and Linux root partitions SATA2 - \users\ on a 1TB 7.2krpm drive, linux home parition SATA4 - 150gb 10krpm Raptor 6gb page file in a 10gb NTFS partition, linux swap2 partition and some other seldom used linux system partitions.

Thus I don't really experience a substantial delay for anything, and re-imaging and backing up is much less painless. I also seldom need to defragment.

You can also use a SSD drive as readyboost cache. Windows 7 allows you to put a ReadyBoost cache file on any drive in the system that passes the performance test.

Re:Forget Heads...

[vadim_t]

It doesn't matter what sizes the writes are, what's important for this case is how many writes per second there are, and you're not giving that number, so I can only guess (the Windows task manager should be able to tell you, btw).

But anyway, that article you posted is completely true. But it's written from a very pessimist perspective, that being that the write limit is easy to reach, so writes must be avoided at all costs. That's a good thing to think of if you're building something for the long term, but most normal hard disks rarely survive more than a few years, and few remain in use for that long either.

As my calculations show, even a 10K erase limit is quite a lot if you spread it over enough sectors.

Intel guarantees their SSDs for 5 years, and modern drives will switch into a special wear reducing mode if you manage to put them under a load that would make them fail too quickly.

Your data is also probably showing the writes the applications make, but not exactly what gets to the disk. There's a write cache, which will eliminate unnecessary writes, and coalesce some of them into one (for instance, writing to a file character by character will NOT result in a disk write per character). And the SSD is free to internally do the same, and take 10 writes to different sectors and put all that data on the same internal block.

Re:Forget Heads...

[RudeIota]

As pointed out above, not all solid state storage is 'flash'. Other technologies such as phase-change memory may have lifetime write cycles in the hundreds of millions per cell. Whether that's a viable technology... well, that's not the point. :)

Every transistor in single cell flash (Flash-based HDD replacements) can be written to hundreds of thousands of times. Firmware logic makes certain to distribute writes evenly across the device.

I'm sure there will be plenty of fallacy in my bad math and short-sighted reasoning, but let's pretend we have a flash SSD that is 64 gigabytes in size. That's about 512,000,000,000 bits (cells/transistors) total.

Also, let's say our flash has a pretty speedy 100MB/sec sustained throughput... And let's make this a perfect world where our transfer rate is always at its sustained peak. That's about 640 seconds to completely use every single transistor on our flash device.

Now, we have to consider that each sector can handle 100,000 writes. 640 seconds x 100,000 times = About 2 years of constant writing at maximum write throughput .

I'd say that sounds just fine for consumer use, considering how extreme the situation is... I wouldn't even expect your average prosumer to have issues with their flash drive for at least the first decade.

Re:Forget Heads...

[rivaldufus]

I doubt you would even have had 50MBs 20 years ago - I sure didn't. I'm thinking more like 10MBs.

as for the old RLL or MFM controllers, maybe someone will release a USB to RLL adapter.

Re:Forget Heads...

[Rockoon]

You missed the part where flash endurance exceeded mechanical drives.

The write limits on modern SSD's is very large. 100,000 cycles and more even on low quality drives. Lets suppose you've got yourself a small 32GB SSD. Thats 32,000,000,000 bytes of capacity, capable of writing at most 3,200,000,000,000,000 bytes.

We can attack this from several angles in order to falsify the notion that this is a hell of a lot of lifetime, but in the end there is simply an upper limit on the number of bytes that can be written per second.

The best SSD's can write about 150,000,000 bytes/sec. Thats 21,333,333 seconds of constant full blast writing. I am quite certain that NO mechanical drive can withstand 247 days of full blast writing. The largest real-world mega-write scenario is probably the daily backup scenario, but SSD's can actualy survive for 100,000 days 9their cycle limit) since you wont be doing more than a full write per day.

For larger SSD's, it looks better and better. A 128GB SSD with a 100,000 cycle limit at 150MB/sec write speed will last for 1000 days of constant writing.

Keep in mind also that 100,000 cycles is on the low end these days.

On TOP of this, SSD's will fail gracefully. You can't write anymore, but you can still read. When a mechanical drive fails, good luck with that.

The modern SSD is hands-down superior to mechanical drives in every way but price and capacity. The mechanicals are now up into the multi-TB range for capacity at only pennies per GB (about 10 cents per GB if you go bargain hunting)

Re:Forget Heads...

[asdf7890]

Did I miss the memo that says flash no longer has a limit on how many times it can be written upon?

No, but the limits are sufficiently high with current technology revisions that it isn't really a problem.

For good solid state drives in all but the most convoluted use cases the expected average time before failure is of about the same order, or some claim better than, spinning disk bases drives. I emphasize the word "good" in that last sentence as this probably may not extent to cheap USB sticks that could be using old design memory and controllers and are generally subject to hasher physical conditions then an internal drive (even in a laptop/netbook).

They key issues with solid state drives at the moment are relative cost (though this will change as the tech matures further), write speed for many small writes (though better drives are coming with more intelligent controllers now, that mitigate this issue somewhat), and write speeds in general particularly after some use (but again, this issue is being actively worked on).

Unless you have a specific use that you think will punish individual flash cells, the write limits should not be a concern when comparing SSDs to spinning disks - instead pick the technology that best fits your desired I/O, power use and noise profiles in your price range.

Re:Forget Heads...

[Barny]

That is a damn good idea.

One thing though, how to redirect /users/ while you mount the drive? (or is there a recovery console command to link a drive into an empty NTFS dir?

Re:Forget Heads...

[ta bu shi da yu]

Look, with wear levelling I really don't think that the Windows registry is going to be a huge concern. I'm even less concerned about MSN Messenger causing the cells to degrade. However, I do agree that for some applications SSDs might not be the best idea.

Re:Forget Heads...

[Rycross]

I'll start welcoming the transition to SSDs when we can get them in a GB/$ ratio greater than 1. SSDs are still terrible in terms of the amount of space you get for the price. The best case scenario is having an SSD for your OS/programs and a rotating disk for mass data storage.

Re:Only one problem with this:

[ProfMobius]

The industry has largely been selling SATA II devices to unwitting consumers based on the perceived promise of 3GBps performance

Well, knowing that the standard is backward compatible (from TFA), what is the point in crying ? You will get a faster interface for the same price as the old one, being able to use your current hardware, and when the drives reach this speed, you will be ready (and from previous posting, looks like SSD are close to saturate SATA II).

Re:Theoretical != Real World speeds

[markringen]

that's going to be too costly, flash is cheaper and easier to produce in huge volume. plus power=unstable, thus flash seems like a smarter choice stability wise.

Re:Only one problem with this:

[macraig]

Yes, I was overlooking the effect of striping multiple drives on the SATA bus, but I doubt if even the fanciest RAID 0 or 5 disk array can come close to saturating even SATA II. SSDs are a much bigger threat, but still pretty costly.

Re:Theoretical != Real World speeds

[hardburn]

SSDs aren't (currently) aiming for the price/GB crown. The power instability is manageable. I'm not saying it's for everyone, but there's definately a niche.

Re:hard drive that can saturate SATA?

[Clover_Kicker]

Prepare for mass storage connected to the north bridge.

/me wanks furiously!

Re:Theoretical != Real World speeds

[Firehed]

Sequential reads on large-capacity drives are often in the 70-90MB/s range (yes MB, not Mb), bursting into the 200MB/s range. Hell, I've seen 50MB/s+ for at least the last half a decade. High-quality (read: expensive) SSDs can roughly double that.

And of course, the spec is in gigabits per second, not gigabytes, and includes overhead. Actual supported, sustained transfer is supported at 150MB/s, 300MB/s, and 600MB/s on SATAI-III respectively.

Re:hard drive that can saturate SATA?

[hardburn]

Agreed that it's eventually going to be on the northbridge. However, SAS isn't there now, either, and SSDs are still likely to saturate that bus in the near future.

SATA vs SAS is a different debate than IDE vs SCSI. Even on servers, it's easy to now justify the cheeper standard compared to the older standards. Not in all cases, of course, but far more often than you could with IDE.

Re:Only one problem with this:

[macraig]

Except the interface specs and other technology will move forward yet again before the devices themselves ever catch up, as has happened with virtually ALL the SATA-bearing motherboards I have ever bought. I'm paying for an interface that I will never be able to fully utilize before the motherboard becomes obsolete e-waste. I don't think the total cumulative combined cost of this interface advancement is as cheap as you think it is, and I don't like paying for something I can't even fully use. I can cite you a far worse example: I once made the misguided choice to buy Pentium-class motherboards with embedded SCSI interfaces. Guess how much I paid for that privilege? $450 EACH, far too big a chunk of it because of the SCSI. I won't do that again, I hope.

My "point in crying", then, is to get other people thinking about the hidden costs and lack of pragmatism involved here. You can disagree or refuse to think about it if you want.

Re:6 Gb/sec? Meh

[The Slashdot 8Ball]

j= - ((-1)^(0.5))

Re:Only one problem with this:

[Firehed]

Hard drives have been shipping with 16MB cache for several years now, and plenty are available with 32MB. RAM is so damn cheap though that they should be seriously considering stuff in the 1GB range for high-end drives.

Re:hard drive that can saturate SATA?

[Kneo24]

Preparing for how much faster your porn will load?

FAQ says $400-500

[c-bo-licious]

That website says $400-500

Re:SSD

[Chabo]

and the mouth breathers that won't get a transponder even though they are now free and clog the automatic lanes

Some of us don't like the government being able to monitor our vehicle's location, and another group of us doesn't like the government to have direct access to our bank account.

Not impressed

[SpitfireSMS]

This is only a margin faster than the new USB 3.0 spec, at 4.9Gbits...

I see more headway being made in the flash storage area.
I really doubt hard drives as we know it will last another couple years.
With SSDs and flash being faster, it only makes sense

Re:Theoretical != Real World speeds

[BlendieOfIndie]

If the two IO benchmarks are *that* far apart, there is something seriously wrong with the test. Mostly likely, one of the tests is reading the same data blocks over and over, and the blocks are being stored in cache (either on the disk, in the array, or on the OS). What type of interface attaches the drives? If it's a 4x SAS connection, then the max theoretical bandwidth is 1200MB/sec.

Beyond that, using the word "sequential" with SSDs makes no sense because the internal bits aren't stored sequentially as they are on a hard disk. Perhaps you meant "read performance".

Nope, system designers want serial comms

[JayBat]

I'd say if it's bandwidth we're after, we shouldn't be reducing the number of signal lines.

Nope, Package pins are expensive, cable connectors are expensive, board traces are expensive, cabling is expensive. On the other hand, silicon is cheap. :-)

A 6gbps serial link is straightforward to implement, if you know what you're doing, and there are probably a couple dozen design groups around the world that can do it.

Jay

Re:hard drive that can saturate SATA?

[afidel]

No mass market SSD can saturate SATA-2 yet, the Intel X-25e at 250MB/s sequential reads is the current performance leader for affordable drives. The next generation will be able to saturate SATA-2 and will probably push SATA-3 before the next SATA spec is finalized if performance trends continue.

Re:Theoretical != Real World speeds

[BikeHelmet]

I really wish SATA 3.0 had a bigger jump than this. 600MB/sec is hardly anything for some of the high end SSDs and RAM-drives available.

If they become affordable, I'm definitely going for PCIe 4x SSDs, since they can hit 8GB/sec (80gbit) when RAID'd on server boards with tons of PCIe lanes.

I remember when someone stuck six FusionIO IODrives together and got about 2.2GB/sec of bandwidth out of a regular 2-socket server board. (like those Tyan ones, which can be had for well under $1000) It seriously makes me drool... though I suppose all I really need out of an SSD is 200MB/sec with minimal latency.

Re:Theoretical != Real World speeds

[afidel]

Actually sequential vs random WRITES make a BIG difference for MLC SSD's and a significant difference for SLC SSD's due to the way that cells are accessed.

I love hard drive technology.....

[ZosX]

Today at work a brand new 1TB seagate came in. I went over to my machine to breathe life back into it to find out that it was instead a 32 megabyte drive according to Windows. Immediately the cache sprang to mind. The drive actually is reporting the cache as the actual drive. Well...hell. At first I thought it was just DOA with corrupt firmware, but after some googling you can actually reset the size that the drive reports with LBA. Hopefully I won't have too many other problems. Not a big fan of the newer seagates, but my boss seems to be going for whatever is cheapest these days.... :/

I would love to get away from complex mechanical drives as a storage medium. Can't someone just make some solid state cube that will hold a petabyte (no petabyte in mozilla's spell checker?? for shame!) and can withstand being written to millions of times?

Re:Theoretical != Real World speeds

[BlendieOfIndie]

Sorry, "two IO benchmarks" was ambiguous - I meant the difference between IOMeter (350MB/sec) and CrystalMark (1504MB/sec). There is an order of magnitude difference between these benchmarks. I suggested that one was affectively measuring cache performance, but I suppose it is possible that one was measuring reads and the other measuring writes. It could also be the number of IO streams hitting the disks. The point is that the numbers posted need to be reexamined.

Re:SSD

[wideBlueSkies]

Yes, but now they don't need the RFID to track your car. Those barcode readers you see at every toll booth make quick work of scanning your registration sticker every time you pass by.

Just sayin'.

Re:SSD

[wideBlueSkies]

ah, this is great. We have the car analogy. Now all we need is for someone to write a post with a Hitler/Nazi reference and we can mark this this one complete.

Re:Theoretical != Real World speeds

[afidel]

A 4x difference between sequential reads and 4K random writes is not only plausible but very likely for MLC, in fact in a non-new MLC it's probably closer to 10x.

Re:3.0

[fuzzyfuzzyfungus]

That's not marketing, that's honesty. The .0 says "Buckle up guys, we are going to be making a whole lot of point releases..."

Re:Theoretical != Real World speeds

[Miseph]

This of course begs the question: other than the ability to say your computer is capable of doing this, what the hell use is it? Are you seriously moving around THAT much data that it is even remotely worth spending the kind of money it would take to actually accomplish this? If your reason really is that it's fricking cool, that's great, but I have trouble believing you have a practical use for this.

Re:Why just double?

[DavidRawling]

That'd be because the SATA specs are coming out every 3 years and the networking ones seem to take 7.

For Ethernet: 10Mbps was 1985, 100Mbps was 1995, 1Gbps was 1999 and 10Gbps was 2006 (ref: Wikipedia 802.3).

SATA went from SATA 1 to SATA 3 between 2003 and 2009 (ref: Wikipedia SATA).

Sorry about the references, CBF doing links today.

Re:Why just double?

[Wesley Felter]

If you increase 10x each generation you have to wait quite a while between generations. You can also end up in a "Goldilocks" situation where 1 Gbps is not enough but 10 Gbps is overkill and too expensive. 2x or 4x per generation is a lot smoother.

Re:Theoretical != Real World speeds

[markringen]

i am betting on ibm racetrack memory. it doesn't matter what the price now is, IBM racetrack memory is around 10 times cheaper to produce (of course IBM is going to charge for it, but it's going to be cheaper than spinning disks)

Speaking of high IOPS and SSD

[symbolset]

Until recently to get decent performance in a reasonable size you needed a huge SAN with hundreds of spindles. Now that you can get stuff like The OCZ Z-Drive, the PhotoFast G-Monster and of course the Fusion-IO IODrive Duo that's not really necessary unless you also need >6TB. The 50 microsecond latency is just bonus.

And oh, joy, there will be more. The SAN vendors who are betting their next year's revenue on those $million+ performance SAN's better get a plan B, and quick.

Re:Speaking of high IOPS and SSD

[Wesley Felter]

SANs typically provide reliable, shared storage; PCI cards don't.

Re:Speaking of high IOPS and SSD

[symbolset]

A SAN is a set of redundant computers with a bunch of direct attached storage and some fancy software. There is nothing magical in that box.

Re:Theoretical != Real World speeds

[ShakaUVM]

>>Are you seriously moving around THAT much data

Fast boot speeds and load times, man, are the holy grail for PC gaming. When SSDs fall enough in price that they're remotely competitive, I'm slapping a SSD RAID0 into my box.

As it is, my 2x7200RPM RAID0 from late 2004 still outperforms a single SSD drive in my SiSoft benchmarks, so I'm happy for now.

Re:hard drive that can saturate SATA?

[symbolset]

Here you go. Knock yourself out - but close the drapes.

Huh?

[symbolset]

There are several SSDs currently that offer more than 1GB/s Read/Write, which would more than saturate this bus. I mentioned them here. The trick is that they don't use this bus. Because that would be silly.

Re:Theoretical != Real World speeds

[symbolset]

What you're looking for is "SATA Expander". It allows you to connect several drives to one 3Gb/s SATA port. The drives, the expander and the controller must all be compatible and support this mode. But yes, you can saturate a single SATA port using only spinning drives if you want to.

Re:SSD

[symbolset]

For random I/O seek time can seriously degrade throughput. Since these days we're talking about multiple VM's contending for disk, this is becoming an issue we need to be more aware of unless we're using Solid State Drives - because of course SSD's don't seek.

Re:Only one problem with this:

[symbolset]

You've missed this one. When CDROM and DVD-ROM manufacturers faced this problem they came up with the obvious answer: Read multiple tracks at once. I believe modern drives read up to 16 tracks at a time.

Only a woman could have

[msimm]

marked your post Informative.

Re:Theoretical != Real World speeds

[BikeHelmet]

Hehe. :)

Well, it would make some of my work go way faster. I imagine creating .ISO files would be whizz bang fast.

But no, I don't have a 24/7 need for it - maybe a 2h/1d need for it. :P It would mostly just be to boast that while most people are stuck at 100MB/sec, my computer is pumping through the gigabytes! ;)

Re:Theoretical != Real World speeds

[evanbd]

Huh? Why would I want to do that when I can just plug each drive into its own port?

Re:Theoretical != Real World speeds

[BikeHelmet]

Hehe. :)

Actually, you may be disappointed if you buy SSDs for the boot speed. HDD manufacturers have done a remarkable job optimizing that. And in the case of an OS like Vista, the benefit from an SSD's low latency is dwarfed by the benefit from an HDD's raw read speed. Those 640GB Caviar drives are apparently one of the fastest booting HDDs/SSDs. (unless you go incredibly high end)

SSDs have been proven enormously helpful for games, though; especially games where you can't possibly store all the textures in memory, such as... Crysis.

Re:Theoretical != Real World speeds

[symbolset]

A few reasons spring to mind. One is that expanders are cheaper than controllers. Another is that they don't take a slot. That's handy if you're using a case that supports 25 drives. A third is that you want to maximize throughput per slot for various reasons. A last is that you want to attach external storage and you want the maximum amount of external storage per connection - because some people want to connect 48TB of storage to one 4-port SATA card, which ain't going to work directly unless you've got a source for 12TB HDDs.

Was that enough reasons?

Re:Theoretical != Real World speeds

[evanbd]

Nope, none of those apply to me.

Only twice as fast?

[Twinbee]

Is it just me or is only a double increase in speed a bit lame (3Gb for SATA 2 to 6Gb for SATA 3).

Re:Only twice as fast?

[Rockoon]

Man I totally agree. They should have jumped right up to PCI-E 3.0 1GB/sec single-lane rates. After all, if even the inexpensive new motherboards are already being equiped with external buses like this, this should neither be expensive nor technically complex.

Re:Theoretical != Real World speeds

[Thing 1]

[...] but I have trouble believing you have a practical use for this.

One acronym: AI (or IA).

I want my computer to tell me what I'm about to think.

Re:6 Gb/sec? Meh

[ion.simon.c]

*Many* people?

Whatevz, you don't fuckin argue with Dr. Emmett Brown.

Re:hard drive that can saturate SATA?

[jhol13]

I think AC meant something like http://www.sun.com/storage/flash/module.jsp.

I do not know how flash is done in laptops now, the best ones cannot be far from the CPU.

Re:Only one problem with this:

[Splab]

Only if they come with a bloody huge sticker telling the customer to make sure this will have battery power. Imagine losing 1 GB of data during power loss....

Re:Theoretical != Real World speeds

[symbolset]

One of the wonderful things about progress is its dimensionality. A rising tide lifts all boats.

So, if you don't see a use for it for you, that's OK. Google and Amazon will find ever more efficient uses for it and so your web pages will load faster and their servers will use less energy and hence burn less carbon so your children will breathe better. People in less advantaged regions will be more able to share their culture and compete in the world economy. The storage clusters I design for your local government will cost 1/10th as much as they did before, so my customers will need less tax money from you.

You win anyway whether you care or not.

Re:Only one problem with this:

[drizek]

Maybe not DISK drives, but we do have SSDs that are approaching 3Gb. SSDs go over 250megs easy, and they are still in their infancy.

We already have SSDs which go well over 3Gb, and they are all PCIe based.

Re:Why just double?

[Cochonou]

SAS also jumped from 3 Gb/s to 6 Gb/s last year. So, it seems to be pretty much the norm for hard drives interfaces.

Re:hard drive that can saturate SATA?

[symbolset]

You need to read up on the literature. And by the literature, I mean the stuff on the shelf at your local online store. Current SSD offerings can beat 1GB/s.

So the difference is the attach. Clearly, this new attach will be useful for people who aren't willing to pay for the premium PCIe attach, or who need volumes greater than 6TB per server.

Re:bike, nigga stole my bike

[porl]

don't worry. by the sound of your attitude towards others it doesn't look like you'll have much of a future anyway. better to give it to someone who deserves it.

Ummm. And the QA checks on the drives?

[Colin Smith]

Are SATA drives as comprehensively tested as SAS? Or have they been selling crap SAS drives with a high markup? I haven't seen any particular problems with SAS, or SATA for that matter, but neither am I running a disk farm.

Re:Theoretical != Real World speeds

[FreakyGreenLeaky]

You mean, like, 640K ought to be enough for anybody?

I can think of several real-world uses in our server room next door... but you were probably referring to him using it on a desktop PC, in which case I probably agree with you.

Then again, if one can afford this kind of performance for a desktop PC, why not?

Saturating current SAS/SATA buses is easy

[Colin Smith]

Any RAID stripe on a reasonable controller and the SAS/SATA bus will at 300MB/s be the I/O bottleneck. Not much point going beyond 4-5 drives at the moment.

What I want though is for 10G ethernet to drop a little in price. Then it'll just be the one technology, and when 10G is too slow for storage I/O, the kit can be reused on the other side of the machine. iSCSI has made FC a legacy technology.

 

Re:Saturating current SAS/SATA buses is easy

[Rockoon]

Its at least a year old now, but look up the "Battleship MTRON" guy who tried to mount like 8 SSD's in RAID0. This was before OCZ and Intel changed the dynamics of the SSD market, and even then he was very near 1GB/sec sustained transfer rates once he found the right RAID controller.

SATAx isnt a RAID controller. While people without good solid RAID controllers can get away with decent RAID0 performance, the serious people never rely on a single SATAx controller for RAID0 since that is not its purpose.

Re:Theoretical != Real World speeds

[Locomorto]

Hehe. :)

Actually, you may be disappointed if you buy SSDs for the boot speed. HDD manufacturers have done a remarkable job optimizing that. And in the case of an OS like Vista, the benefit from an SSD's low latency is dwarfed by the benefit from an HDD's raw read speed. Those 640GB Caviar drives are apparently one of the fastest booting HDDs/SSDs. (unless you go incredibly high end)

SSDs have been proven enormously helpful for games, though; especially games where you can't possibly store all the textures in memory, such as... Crysis.

I think your post is either terribly misinformed, or you are overstating yourself. I have no idea how can say a 320GB platter drive (and there are 500GB ones now too...) could possibly hope to outperform all but 'incredibly high end' drives. Look at the OCZ Vertex, or G.Skill Falcon (or the supertalent drives etc based on the same controller). The smaller sizes are around the price of a velciraptor (admittedly smaller though, but still large enough for vista + some apps), and they blow the velciraptor out of the water. Don't just take my word for it: http://www.overclock3d.net/reviews.php?/storage/ocz_vertex_120gb_sata2_ssd/7 And that crap about it being enormously helpful? Sure it definitely is desirable, but not even that benchmark you posted is fair for crysis (as they admit in the article...). For the vast majority of the time, SSDs only improve load time (note your floppy RAID array does not count).

Re:Theoretical != Real World speeds

[bertok]

I got an OCZ VERTEX SSD drive, and it outperforms a SAN with a whole tray of disks for real-world apps, so you may want to check out the latest stuff. You have to be careful with benchmarks, they all measure throughput of heavily pipelined IOs, which almost never happens in real applications. A mechanical disk, even in a RAID, will struggle to outperform SSDs that can do reads in well under a millisecond.

Re:Theoretical != Real World speeds

[petermgreen]

One is that expanders are cheaper than controllers.
When I looked the per port cost of controllers seemed pretty similar to the per port cost of multiplier capable controller with multipliers.

Another is that they don't take a slot. That's handy if you're using a case that supports 25 drives.
You still have to mount them somewhere though, most i've seen seem to be intended for special mountings found in dedicated external storage enclosures. I've also seen expansion slot mountings but that kinda defeats the object of saving expansion slots.

A last is that you want to attach external storage
This seems to be the main attraction to me, you can have an external enclosure with 4-10 drives and only one or two cables connecting it to the main machine.

Re:Awesome

[petermgreen]

What kind of drive is this? unless it is a SSD I highly doubt the SATA connection is the bottleneck.

Re:Only one problem with this:

[petermgreen]

You've missed this one. When CDROM and DVD-ROM manufacturers faced this problem they came up with the obvious answer: Read multiple tracks at once.
They certainly tried it, I don't remember it being much of a success though. They used some optical tricks though rather than multiple read heads.

The problem with reading multiple tracks at once on a hard drive is you would need multiple head assemblies and control circuits for them. This would put up the price hugely and decrease reliability.

Re:Only one problem with this:

[mikael]

Easily happens with CPU systems - nothing would piss off a technical writer, animator, 3D modeler or texture artist more than having their workstation crash after several hours of editing. Either caused by a electrical fault or a software bug.

Fortunately, there are Battery backed RAM disks. Thought these had died
out after all these years.

Re:Theoretical != Real World speeds

[Guspaz]

Your 7200RPM drives probably have a sequential transfer average of something like 50 or 60 MB/s each, for 100 to 120 MB/s total. A mainstream Intel SSD does 230MB/s average (using my own drive's benchmarks). As for random read/write performance, you'd need dozens (or more?) of drives in an array to match up a single SSD.

Ignoring write performance (where the Intel drives still slaughter a traditional HDD in random performance, but fall a bit behind in sustained), you're still far behind an SSD.

Re:Theoretical != Real World speeds

[evanbd]

Oh, I'm well aware of how progress works and that other people might have a use for them. I was just pointing out that your post:

What you're looking for is "SATA Expander".

was inaccurate. I am not looking for any such thing. If I want more performance, I'll get more disks or faster disks, and I'll attach each one to its own port. I have extras.

Spec should include the following for MOBO makers

[motherpusbucket]

"Thou shalt keep SATA connectors the hell away from PCI-x slots." It amazes me how many MB's still put them in locations that interfere with vid cards.

Re:SSD

[Chabo]

What state do you live in? I don't have a barcode anywhere on my car.

Regardless, even if they can track my car now, I still don't like the idea of giving the government direct access to my bank account.

Bandwidth isn't really the problem

[Leolo]

I don't really care about bandwidth. What I really care about is parallel requests and out of order requests. This is why SCSI was so much better then IDE. Does SATA 3.0 remove the odious limit of 15 NCQ reqs?

Not fast enough I suspect

[BloodyIron]

6Gb/s is pretty good, but I suspect as SSD's accelerate in development that the 6Gb/s limit will be reached before the next generation of SATA. How long has SATA 3Gb/s been around? Quite a while...

Re:Theoretical != Real World speeds

[Miseph]

It seemed from his post that this was for a desktop PC, and not a high use DB server or some such. And yes, to a certain extent, 640k WAS enough for anybody. Those statements shouldn't have to come with an explicit expiration date. At this point in time, there are extremely few uses for that kind of bandwidth, and the foreseeable future doesn't hold many more.

Re:Theoretical != Real World speeds

[ShakaUVM]

Like I said, I ran a series of benchmarks on my drive, and they scored higher than the (downloaded) SSDs in the system.

Re:Theoretical != Real World speeds

[ShakaUVM]

>>AHAHAHAHAHA

After FPS of course. And load speeds make a big impact on real life (i.e. not benchmarked) gameplay. You wander in Crysis to a new part of the island at high speed, the game doesn't have time to page everything in, and the game gets real choppy, real fast, regardless of your 10xSLI video array that needs to be powered by a portable nuclear reactor.

That and the fact that waiting a couple minutes for a level to load kills a lot of the fun.

Re:SSD

[wideBlueSkies]

New York.

The Empire State.