TB-Sized Solid State Drives Announced

prostoalex writes "Several companies have announced solid state hard drives in excess of one terrabyte in size. ComputerWorld describes one from BitMicro that's just 3.5". Their flash drive will support up to 4 Gbps data transfer rate. From the article: 'SSDs access data in microseconds, instead of the millliseconds that traditional hard drives use to retrieve data. The BitMicro E-Disk Altima 4Gb FC delivers more than 55,000 I/O operations per second (IOPS) and has a sustained data transfer rate over 230MB/sec. By comparison, a fast hard drive for example will run at around 300 IOPS.'" Ah, the speed of tech. Seems like only last month we were talking about 500GB drives.

Comments

the bus will be the bottleneck

[petes_PoV]

To get the best performance out of these things we need to move away from IDE/SATA architectures and have the storage directly on the PCI or PCI-E bus.

Once that happens, PCs will really start to get useful!

Re:the bus will be the bottleneck

[Kjella]

Right now, even the best prosumer SSDs aren't close to matching the SATA2 interface speed of 300MB/s. Their access time is already minimal (I saw 0.2ms in a test for one disk). Remember that there's a lot of headroom in the SATA interface that isn't used except for buffer bursts today. So I don't really see the point. Having an standard connector and making them drop-in replacements for HDDs sounds to me like the logical way to go...

Very nice

[Bullfish]

But for now the cost isn't worth the performance differential. With enough ram, generally you aren't hitting the hard drive too often except for a few tasks. With 64 bit computing, you get to have even more useful ram. When the price of solid state drives is competitive with hard disks, I'll pay attention.

Re:Very nice

[itsdapead]

When the price of solid state drives is competitive with hard disks, I'll pay attention.

Hah. When I were a lad you could get a 7 MEGABYTE Winchester Hard Disk for a mere £3500 (what, about $5000?). (Source, 1981 copy of Personal Computer World).

That's about £10k in modern money (according to this calculator - a.k.a. $20k dollars (or $10k at Microsoft/Adobe screw-the-Brits rates).

Now, if you think that 1981 was, like, ancient history then GET OFF MY LAWN! If the usual growth rate applies, 1TB solid state drives will be cheap and plentiful by the time you get round to repainting your house.

Re:Don't Forget Fusion IO's PCIe Card Drive

[Junta]

The Texas Memory Systems datasheet claims 24 GB/second of random sustainable data bandwidth which is much higher than the Fusion IO card but it looks like they are serializing this possibly across multiple drives. They also claim higher (3.2 million) operations per second. The Texas memory product is cramming a bunch of ram in 24U and putting a whopping 25 minutes of battery backup. If you disconnect power for more thon 25 minutes, the only thing left is whatever was committed to 'real' peristant storage. They provide Infiniband and FC ports, so it's more akin to an EMC or Engenio storage controller than a hard drive. 1 TB/24U is actually kind of sad when hard drives can easily yield 3 TB/U nowadays. There is a place for this (ramdisk performance is pretty nifty), but it's not even remotely relevant to anything a normal person would think of when they hear SSD (they picture a drive intended to connect directly to a system some how, not participate in a SAN directly.

The BitMicro drive is groin grabbingly amazing in size but claims only 55k operations per second & sustained data transfer rate over 230MB/sec. And it *actually* is flash based storage, meaning it can fairly be called persistant storage. Of course, the clear hint is there when they talk IOps and only mention FC connection that they are targetting only deep pockets with the product as of this press release or whatever. 55k operations and 230 MB/s is ludicrously insane performance for a single drive relative to current spinning disks. You can fit 144 of these into a 24U space and have a theoretical aggregation that exceeds the ram based system specs. Of course, RAM should be able to trounce it so the limiting factor is a controller setup to push the IOPs and throughput, so both solutions would probably perform comprably.

So what I would wager is that PCIe might provide more throughput than SATA but don't quote me on that. I'm interested to see where this goes & also curious to see whether we continue dumping drives on channels like the Texas Memory solution or if it just goes back to a server with a ton of PCIe slots on it and hot pluggable card swapping for 'drives.' Well, considering that SATA controllers at best currently use PCI-e as the method to communicate with the chipset, PCI-e slots better be capable of better than SATA... Ok, it's over-simplyfying, a 1x PCIe first gen slot yields about 2.5 Gb/s or so, and a SATA II port is 3.0 Gb/s, so a single lane PCI-e slot would be slower than a SATA port. However commonly PCIe appears in 4/8/16 lane configurations, I assume you meant 880 MB/s, which would point to PCIe 4x slot sort of throughput, which makes sense, it's not unreasonable to expect at least a 4x lane slot to be free, requiring anything more could waste limited hardware resources.

Re:ReadyBoost, et al

[mikiN]

Interesting idea, but somehow I can't escape thinking this is like mashing up an iPod in a blender because the resulting grey powder looks nice in the mock-up fireplace for your Christmas stall.

Unless there have been some really important changes in the performance of Flash memory, using it as swap would be like the second worst possible scenario in terms of it's life expectancy (using it for main memory would be the worst). Just how long is a typical Flash chip with a guaranteed average of 1 million write cycles going to last in this kind of environment? Hours? Days? Perhaps even...weeks?

Re:Quick Erase?

[Jherek Carnelian]

Well, why would you need to erase your entire drive for anyone besides the police? Well, why would you need to encrypt your entire drive for anyone besides the police?

A laptop with a "quick erase" button would be useful to many businesses with people who travel in less than the safe regions. Same goes for people in less than free countries who are working for change. Full disk encryption doesn't stand up to rubber-hose cryptography very well.

I'm sure with more than 30 seconds to think about I could come up with any number of 'legitimate' uses.

Filesystems

[peterlombardo]

Since filesystems are so closely tied to cylinders, tracks, sectors and blocks...how does this play on SSDs? If I'm not mistaken, when allocating new extents, filesystems take into account physical locations to minimize future seek times...is that valid on a SSD?

Not Yet Convinced

[Nom du Keyboard]

I'm not yet convinced that paying a premium price for a hard drive using a more expensive technology with a very (compared to rotating storage) limited lifetime in terms of write cycles is a wise idea. There are parts of my hard drive (swap areas) that get beat-up pretty badly at times. Don't want to wear this thing out in a year or two.

That's also why I don't have a plasma big screen yet. I'm using an alternative technology there as well.

Re:Quick Erase?

[dossen]

Just to pick nits - the commonly used term is rubber-hose cryptanalysis (http://en.wikipedia.org/wiki/Rubber-hose_cryptanalysis).