RAM Disk Puts New Spin On the SSD

theraindog writes "Although the solid-state storage market is currently dominated by flash-based devices, you can also build an SSD out of standard system memory modules. Hardware-based RAM disks tend to be prohibitively expensive, but ACard has built an affordable one that supports up to 64GB of standard DDR2 memory and features dual Serial ATA ports to improve performance with RAID configurations. And it's driver-free and OS-independent, too. The Tech Report's in-depth review of the ANS-9010 RAM disk pits it against the fastest SSDs around and nicely illustrates the drive's staggering performance potential with multitasking and multi-user loads. However, it also highlights the device's shortcomings, including the fact that SSDs are more practical for most applications."

Comments

What I learned from the article

[Wonko the Sane]

• Most tasks are not disk IO-bound
• Despite the fact that this device uses DDR2 RAM running at more than 6 GB/sec, it can not saturate 2 SATA interfaces
• Why bother?

Re:What I learned from the article

[NevarMore]

I'd bother, because most of *my* tasks are disk I/O bound.

Re:What I learned from the article

[Wonko the Sane]

But are you really getting your money's worth from this device?

DDR2 is an order of magnitude faster than SATA. Looking at their numbers, the internal controller is limited to about 400 MB/sec. That is pretty mediocre.

Re:What I learned from the article

[Poltras]

Though the motherboard that will support 64Gb of RAM cost more than this baby here. And we're not talking about the OS to support those. In a server environment, I can see a use for faster disks when memory is maxed out.

Re:What I learned from the article

[mariushm]

The reason it can't saturate the 2 SATA interfaces is most likely the custom chip they use, probably a Xilinx FPGA chip like the one i-Ram used.

Had they invested lots of cash in making a custom chip (but this takes time, months), it may have been faster but I assume they either believed it won't be a success, or got greedy and wanted to put something on the market fast or they assumed it's fast enough to make some fast cash.

Right now the only problem I see is the short 4 hour battery time and maybe the low transfer speed to CF cards (it needs about 20 minutes to transfer 16GB to backup card).

They should supply this device with a USB or PS2 cable adapter thingy, because most BIOSES are configured by default to power keyboards/mouse devices and some usb ports (or can easily be configured to do so and power draw is not so much - 12W is probably too much for that adapter but the device doesn't use that much when computer is turned off).

Re:What I learned from the article

[Vellmont]

In a server environment, I can see a use for faster disks when memory is maxed out.

I can see it, but the use is fairly limited. The usage would have to be an environment where you need both extreme performance, and don't care about losing the data on the device. (The battery is only good for 4 hours, not a terribly long amount of time).

Those situations exist, but are relatively specialized.

Re:What I learned from the article

[Glonoinha]

it needs about 20 minutes to transfer 16GB to backup card

That's 16 megabytes per second - if I had to guess, the bottleneck is the CF transfer rate and has nothing to do with the rest of the device.

Re:What I learned from the article

[Forge]

This brings up an interesting idea.

What if the ramdisk function was moved into the motherboard chipset? This would achieve 2 things:

1. It would dramatically cut the cost of a ramdisk. I.e. The cost of the entire motherboard might go up by $5 or so.

2. It would eliminate that SATA bottleneck, allowing ramdisk to run at full RAM speed.

If you then figure out a way to have this data loaded to the ramdisk from a hard drive at poweron (or get realy clever and mount a flash chip on or near each DIMM which takes a backup of that DIMM, just before powerdown.

Re:What I learned from the article

[AmiMoJo]

You can get SSD cards with a PCI-e interface that hit 800MB/sec. Why RAM disk manufacturers stick to SATA I don't know.

PCI-e even has from standby power available.

Re:What I learned from the article

[Wonko the Sane]

You're confusing bits and bytes.

Re:What I learned from the article

[LWATCDR]

I don't see a huge problem with the battery backup time or the transfer speed.
I would use this in a database server. Possibly even RAID them for even more speed. If the system is without power for four hours then there are some REAL problems going on.
It would also be good for a rugged embedded systems/
The speed will come up and Ram prices will keep dropping.

Re:What I learned from the article

[Ed Avis]

Why would you spend your money on this device instead of just buying the equivalent amount of RAM and putting it on the motherboard where the processor can access it directly? Even if you had to upgrade to a more expensive motherboard you'd still get way better price-performance by doing that, rather than crippling the RAM by putting it on the other side of the SATA bottleneck.

If you insist on having a 'disk' you can save files to, well, all OSes support the idea of a RAM disk...

Re:What I learned from the article

[powerlord]

You can get SSD cards with a PCI-e interface that hit 800MB/sec. Why RAM disk manufacturers stick to SATA I don't know.

PCI-e even has from standby power available.

Well ... I would imagine that the main reason they stick to SATA is the ubiquity of the interface.

At this point, just about every motherboard or computer out there contains a SATA drive, even if it doesn't have a PCI-e slot (or a slot available, or a slot of the required bandwidth available). That, theoretically, expands their potential market.

Re:What I learned from the article

[drsmithy]

How does that NOT saturate 2 SATA II interfaces?Because they've cheaped-out on their magic RAM2DISK chipset and it apparently maxes out at ~400MB/sec.

Which is, as another poster mentions, a bit crap considering that the onboard RAID controller on a fairly mundane server (eg: Dell PE2950) can handle higher throughput.

Re:What I learned from the article

[mcrbids]

This brings up an interesting idea. What if the ramdisk function was moved into the motherboard chipset?

OMFG! That's an AMAZING idea! This could dramatically change computing as we know it! The implications of this are, eh, well....
 
.... quite well understood. Somebody thought of this many years ago. Many, many, many years ago. It's called a (ahem) "ram disk" and uses system memory as if it were a drive with a software driver. Here's a howto for Linux - I did something similar with so-called "high memory" on a 80286 with DOS 3.x and ramdrive.sys - that 384k ram disk was small, but //FAST//!!!

Sorry to break the news to you.

Re:What I learned from the article

[conureman]

This looks like an excellent solution for a sound mixer/cd burner box I'm building for a musician friend, studio-quiet and fast. I'm hoping to wow him with the upgrade, from a generic P4 box to an Opteron. Should be a noticeable improvement. Maybe even an ANS9010B, as my friend is typically cheap.

Re:What I learned from the article

[Forge]

Doesn't everyone play with driver level software ramdisks at some point?

What I was talking about is to move that software function out of the OS entirely so that the computer could boot off a RamDisk.

If done right, you would simply have an insanely fast hard drive for OS, Swap and your favorite game. Leaving that 8.5 TB RAID 6 for your porn archive.

Re:What I learned from the article

[dotancohen]

This brings up an interesting idea.

The real question that is not being answered here is why does a 32 GB SD card cost $25 but a 64 GB SATA hard drive cost $800? Why can't the technology that makes SD cards so cheap make cheap SATA hard drives as well?

If I'm missing something obvious and this sounds like a troll, then please RTFM me with a link, because I'd really like to know.

Re:What I learned from the article

[clarkn0va]

Shorter boot time. With that one exception I think you make a solid point.

Re:What I learned from the article

[billcopc]

Okay, as a system builder and all-around hardware nut, allow me to translate your bullet points into plain english.

* Most tasks are not disk IO-bound

Define "most". Consumer systems are very IO-bound. Office apps load a gazillion files, games read and write large data lumps with miserably inadequate buffers (causing massive seeking and cache inefficiency), media apps read and write large video files and usually have massive intermediate files while encoding/rendering. Web and database servers are almost always IO-bound under heavy loads.

        * Despite the fact that this device uses DDR2 RAM running at more than 6 GB/sec, it can not saturate 2 SATA interfaces

That's because this interface is cheap and made by a Taiwanese gadget company, known for cheap knockoff products, not high-performance equipment. The whole idea of using SATA is kludgey at best, when you could be using the full potential of the PCI-Express bus.

        * Why bother?

Because cheap Taiwanese knockoff companies need money too, and their not-quite-useful products always appeal to the techno-suckers. If this product were a serious Ram-disk, it would cost ten times more and wouldn't be hung from a hook on a pegboard at your local computer scamshop.

Re:What I learned from the article

[42forty-two42]

This is what is often referred to as a "disk cache", and has been implemented in all "real OSes" since the 70s :)

Re:What I learned from the article

[billcopc]

If a motherboard maker is going to waste effort on an onboard Ramdisk, I'd rather just have them put two dozen memory slots and let me install a couple hundred GB of DDR2/3 for a 64-bit OS to use.

What you have described is an old sloppy hack. It's something guys like me used to do before 64-bit PCs became affordable. 4gb Ram + 4gb swapfile on a Ramdisk = ghetto 8gb heap.

The real solution to the disk/Ram problem is to rethink how we use these devices. We can't keep on growing our operating systems to fill all available memory, that won't be sustainable for much longer as performance levels are worse than they were ten years ago, despite the hardware being 10 times faster across the board. Eventually we will hit a wall, and all those sloppy coders will have to shape up or start driving buses.

Re:What I learned from the article

[Wonko the Sane]

Define "most"

Most of the applications that were benchmarked in the article.

That's because this interface is cheap and made by a Taiwanese gadget company, known for cheap knockoff products

Exactly

Because cheap Taiwanese knockoff companies need money too, and their not-quite-useful products always appeal to the techno-suckers. If this product were a serious Ram-disk, it would cost ten times more and wouldn't be hung from a hook on a pegboard at your local computer scamshop.

My point was that this product is a toy. If you really need a ramdisk, get a real one.

Re:What I learned from the article

[Anubis350]

To be fair, I've worked with using ramdisks exported over highspeed interconnects (like inifiniband) as netboot disks, very often the server+ramdisk+interconnect (notice the lack of drive connection there) is faster than a local disk, so moving this into hardware for boot support would be *very* nice.

Re:What I learned from the article

[severoon]

Maybe most tasks are not disk IO-bound—but when the few that are burn 95%+ of the time you sit at your computer waiting, that makes TFA relevant.

I do a lot of heavy post-processing in Photoshop (thank god I'm not into video). I need a 64-bit machine that will allow me to start with 8GB of available RAM and has an on-board SS disk cache. They've already started doing this with laptops. Wouldn't it be great if when you shut down your machine, it automatically rebooted and ripped the memory image of your freshly booted machine into a SS cache? Next time you poke the power button, in a few seconds it could rip the image bit-for-bit into RAM and you're ready to start computin'.

And it would be awesome if I could designate parts of the HD to cache as well. I would like to use a RAM disk, but Photoshop requires its swap file be placed on the same drive as the application itself...so why not just make the SS cache transparent via the OS? Any files experiencing heavy load get cached automatically in SS, like this Photoshop swap file, the Windows swap file, etc.

I expected these motherboards the day came the first 8GB CompactFlash cards were released. What's taking so long???

Re:What I learned from the article

[dougisfunny]

I think SD cards are a lot slower.

Re:What I learned from the article

[X0563511]

This isn't RAM in the classical sense. It's a storage device that uses RAM instead of a magnetic platter.

As long as your motherboard supports SATA, then it supports this.

Re:What I learned from the article

[mrchaotica]

why does a 32 GB SD card cost $25 but a 64 GB SATA hard drive cost $800?

I was wondering that myself. I've got two guesses:

1. Perhaps the flash 'hard drives' are using really fast flash memory (like the kind professional photographers use, where they need to be able to shoot 10 megapixel photos at 10 FPS).
2. Perhaps they were just gouging, and prices have now dropped as more have become available (for example, compare: $75, $577).

Re:What I learned from the article

[X0563511]

Except that the data still has to go through the southbridge, that will be your new bottleneck. RAM is through the northbridge, which is loads faster... but is also directly tied with the CPU(s).

Re:What I learned from the article

[X0563511]

What use is it to boot from a ramdisk like that? The CPU cannot execute zeroes.

Re:What I learned from the article

[Forge]

Everone on SlashDot knows "More's Law": Every 18 months CPU speed doubles. Most also know "Gate's Law": Every 16 months the speed of software halves. The gap in those timespans is why machines keep getting slower.

Re:What I learned from the article

[Chabo]

Most tasks are not disk IO-bound

I beg to differ. I have a P4 3.2, 1GB DDR400, and a WD640 Black. So far as I know, the WD640 Black is still the fastest mechanical drive on the market, but it's still the bottleneck for most tasks I run. Level loads in games, CD/DVD ripping and burning, and Windows' boot all are disk-bound for me, and my system's not all that great. Ok, video encoding may be CPU-bound for me, but audio encoding is fairly evenly matched (it's CPU-bound, but the disk is working pretty hard), and I spend far more time ripping and burning CDs and DVDs than encoding them to other formats.

Re:What I learned from the article

[Chabo]

Let me correct myself:

So far as I know, the WD640 Black is still the fastest 7200rpm drive on the market

Re:What I learned from the article

[SanityInAnarchy]

I don't really see how this would beat an SSD or a NAS-in-another-room for the same purpose. Mixing/burning doesn't exactly take a lot of disk IO, does it?

Re:What I learned from the article

[SanityInAnarchy]

What I was talking about is to move that software function out of the OS entirely so that the computer could boot off a RamDisk.

So, presumably either there's a persistent store to back it, meaning you're just booting from that RAID anyway...

Or it's got some amount of power running through it, maybe backed by a battery. ...sounds pretty much exactly like a laptop in suspend-to-RAM. Otherwise known as Standby, or Sleep. If it's "safe sleep", you're even safe if power cuts out completely.

Re:What I learned from the article

[Seq]

Cache/swap for video or large image editing?

Assuming one does not have a motherboard that supports crazy amounts of memory, or for retrofitting into existing machines (such as a mac pro)

Re:What I learned from the article

[Wonko the Sane]

I'm speaking in the context of the article. Most of their application benchmarks showed only a minor speed difference between this drive and regular hard drives.

Re:What I learned from the article

[amoeba1911]

I'd like to add to that, the best place I can think of for putting an SSD would be a laptop since SSD:
has lower energy consumption
is faster than any physical hard drive you can get for a laptop
has less heat to dissipate
is immune to physical abuse

Those are the qualities you want in a laptop hard drive and SSD has all of them. I guess that's why most (or all) SSD's come in 2.5" size (for laptops) instead of the more prevalent standard desktop hard drive size of 3.5" or other standards that are specific to desktop machines. ... they belong in laptops. :)

Re:What I learned from the article

[PitaBred]

My 16GB SDHC is class 6, which means it'll write at 6MB/s. That's significantly slower than an SSD drive. It is because of the difference between SLC and MLC types of flash, really.

Re:What I learned from the article

[default luser]

The real question that is not being answered here is why does a 32 GB SD card cost $25 but a 64 GB SATA hard drive cost $800?

A 32GB SDHC card right now on Newegg (in-stock) is a minimum of $72. I don't know where you got the $25 number (sure, in another year it will be that cheap). As another poster mentioned, Newegg has 32GB SSDs available for the same price range.

Why can't the technology that makes SD cards so cheap make cheap SATA hard drives as well?

It already has. The first SSDs on the market used single-level cell (SLC) flash, while the inexpensive SD cards and mp3 players you see everywhere use multi-level cell (MLC) flash. The difference is how densely you can pack the data, and it makes a huge difference in price.

To put it simply: SLC flash is faster, lower-power, and more reliable than MLC flash, but also more expensive (at same capacity) than MLC flash.

The reason the first SSDs used SLC flash is because new technologies have to convince people to take the plunge: people/companies are usually willing to pay significantly more for something that is much faster and more reliable. Early adpoters might have given SSDs the cold shoulder if the first wave of drives reduced capacity and performance in-order to be more cost-competitive with existing storage.

Now that SSDs are firmly off-the-ground, manufacturers are offering all sorts of devices, including cut-rate drives using MLC flash, so the prices at the low-end have dropped like a rock.

Re:What I learned from the article

[sirsnork]

It wouldn't eliminate the SATA problem at all. The OS still has to have a driver to talk to the "disk". If it's not SATA then what? I suppose you could use SAS but that doesn't really buy you anymore speed at this point.

The problem with this benchmark overall seems to be that their SATA controllers (both the one on the motherboard and the one in the drive) can't do anything like it's ~300MB/sec per port. This is glaringly obvious in the file copy tests where the extra port gives you a boost but nothing like twice the speed and even on one port they are still a long way from ~300MB/sec.

Of course in everyone's defense they haven't really ever been able to test this in the past without some custom hardware.

Re:What I learned from the article

[hairyfeet]

Do you have WinXP? Do you have plenty of RAM in your machine? Then save yourself some cash and use a free RAM disk in XP. Just go here and they have the step by step. I have used one and I only have 2Gb of RAM(which tickles me to say since I started on a VIC with what? 1Kb of RAM?) and for I/O intensive tasks it does give it a good kick in the butt. It is also a GREAT way to get around that XP 32bit 3.5Gb RAM limit. I set up a gamer rig that had 6Gb and XP 32bit with a persistent 3Gb RAM disk and all his friends are amazed and what a screamin demon of a machine he has thanks to using the RAM disk as a temp file. It is also nice for security, since you can have your browser use it as a temp folder and everything goes poof on reboot. Try it, you'll like it.

Re:What I learned from the article

[dotancohen]

Wow, that is quite a wide price range.

Re:What I learned from the article

[ColdWetDog]

Why don't you get a MacPro? You can have up to 32 GB RAM, you can suspend to RAM in seconds. And restart successfully almost all of the time....

I'm doing multi gigabyte panoramas in PS (and some video encoding). The MacPro is great for this. And when I'm bored, I put it into suspend (mostly to power down the monitors) and come back later. No worries.

Come on in, the Kool-Aid's fine.

Re:What I learned from the article

[bill_mcgonigle]

There's always a limit to how much RAM you can put on a motherboard. Always!

Sure, but is it a practical limit? I can easily buy 128GB of RAM in a server. For a bunch of money I can buy 256GB of RAM for a server. And I get hardware ECC and FSB bandwidth.

These only have 8 slots, so given your insight, these are only really practical if you need more than 128GB of RAM-disk in a server or need the (preposterously slow) store-to-CF feature. Or your server is crashy.

Is anybody here using a write-mostly RAID with a linux RAM-disk or ZFS?

Re:What I learned from the article

[Score Whore]

Put them into a Solaris ZFS pool as the logging device or the l2arc. For working set sizes in the area of fifty or sixty gigabytes, it'll make your three or four terabytes across ten spindles act like three or four terabytes across one hundred spindles. Sounds worth it to me.

Re:What I learned from the article

[adisakp]

Despite the fact that this device uses DDR2 RAM running at more than 6 GB/sec, it can not saturate 2 SATA interfaces

What would be much smarter is if someone came out with a PCI express card that took RAM slots so it could provide a RAM disk with 6GB/sec throughput instead of the limited ~600 MB/s (SATA 3GB/s X 2 RAID0). On top of that, add multiple memory buses (on card RAID again) to the DIMMs so it could hit the theoretical 16GB/s you get with 16 lanes of PCIe 3.0.

Re:What I learned from the article

[CthulhuDreamer]

I used to have an old PowerMac 9600 with 12 memory slots using 64MB or 128MB sticks. Since the OS at the time could only use 512MB, any extra RAM could be split off into a bootable ramdisk. With the OS and my applications running off the ramdisk, that clunky 300MHz processor seemed faster than anything I've able to buy since. It seems we've been going backwards in some aspects.

Re:What I learned from the article

[AK Marc]

It wouldn't eliminate the SATA problem at all. The OS still has to have a driver to talk to the "disk". If it's not SATA then what?

Make it PCIe and write a simple driver based on SATA or SAS or whatever.

The problem with this benchmark overall seems to be that their SATA controllers (both the one on the motherboard and the one in the drive) can't do anything like it's ~300MB/sec per port.

The real win would be having it at near-RAM speeds, rather than fast disk speeds. OSs can handle ramdisks, so have it be a ramdisk that has the ram physically separate on the MB with a small battery or supercap for storage when powered off overnight.

Of course in everyone's defense they haven't really ever been able to test this in the past without some custom hardware.

And anything like this would only be done with custom hardware. The trick is to make it work, work well, and then others will adopt the custom hardware and make it a standard. Even better if you can sneak in a couple patents and trademarks, hand it out for free, wait 10 years, and sue everyone.

Re:What I learned from the article

[AmiMoJo]

At this point, just about every motherboard or computer out there contains a SATA drive, even if it doesn't have a PCI-e slot

Are you seriously suggesting that someone looking to spend hundreds of pounds on a RAM disk won't have a PCI-e mobo? Even the cheapest mobos have PCI-e now, and you can fit 4x cards into both 16x and 1x slots.

Re:What I learned from the article

[pclminion]

That's not what I want. I don't want to have to buy the fastest, most expensive RAM available just to use as a RAM disk. I'd prefer cheaper RAM, maybe two or three generations old, that I can get in massive quantities. Sure, put it on the motherboard, but don't put it on the main memory bus -- that's not its purpose, and I don't want it eating up gobs of the normal address space. It's for storage only.

It doesn't have to be as fast as main memory, just faster than your typical disk interface. Make up an entirely new interface if you want. Call it RID -- RAM integrated drive.

Re:What I learned from the article

[sigismond0]

And here's why SSD's are good for desktops: Lower energy consumption Faster than any physical drive you can get for a desktop Less heat to dissipate Good job. SSD's belong everywhere. Well, except for heavy I/O tasks.

Re:What I learned from the article

[Mozk]

2 Gb is only 256 MB. There is a difference between Gb and GB.

And I'll just ignore that you didn't put a space between the numbers and their units and that technically this should all be MiB and GiB.

Re:What I learned from the article

[Monkeybaister]

The very simple answer:

The motherboard is already at the limits and you still need more.

or

You're dealing with some application programmed by cavemen that insists on having a disk (or swap) to write things to and the clever (yet still moronic) programmers decided to check whether you are trying to cheat. I've delt with some pretty shitty programs, so I don't doubt it's happened.

Re:What I learned from the article

[Ken_g6]

The site says that ramdisk is limited to 32MB! How'd you get a 3GB disk?

There's also mention that it may not work with, or may cause problems with the System Restore feature in XP and Vista.

Re:What I learned from the article

[powerlord]

A 4x PCIe card will not fit into a 1x PCIe slot.

Go learn more about the spec. "A PCIe card will fit into a slot of its physical size or bigger, but may not fit into a smaller PCIe slot. Some slots use open-ended sockets to permit physically longer cards and will negotiate the best available electrical connection. The number of lanes actually connected to a slot may also be less than the number supported by the physical slot size. An example is a x8 slot that actually only runs at x1; these slots will allow any x1, x2, x4 or x8 card to be used, though only running at the x1 speed. This type of socket is described as a "x8 (x1 mode)" slot, meaning it physically accepts up to x8 cards but only runs at x1 speed. The advantage gained is that a larger range of PCIe cards can still be used without requiring the motherboard hardware to support the full transfer rate - in so doing keeping design and implementation costs down." (from the requisite Wikipedia page)

As for motherboards without a PCI-e slot:

1) some 1U servers. Granted most of them seem to have at least one slot, but then you can get into issues with how much clearance the actual card needs and cable runs (which rack mounted servers can be tricky about, but is not relevant to the discussion we're having). You might also have issues with Blade servers using an expansion card to power adjacent storage blades. Its very doable, but it might not be your first choice. Dropping disks directly into storage blades might be "easier" from a logistics and support perspective, which can be a huge benefit to a lot of larger companies.

2) laptops/portable devices. There is a reason SSD drives are coming in 2.5" hard-drive form factors. The theoretical combination of lower power requirements, quicker response, more resistance to jostles and bumps, make it SSDs a good fit.

Yeah, I don't know what the power requirements of the RAM disk are versus an SSD, but if they are competing against SSDs, then it is a market segment they may also be targeting.

Re:What I learned from the article

[longbot]

I think you're missing the point here.

Software RAMdisks have been around for ages. What Forge is proposing is moving the interface function that's done by the device in TFA onto a a chip adjacent or integrated with the memory controller on a PC motherboard. That would be a pretty revolutionary idea.

I imagine a cluster of machines doing this, all interconnected with fiberchannel, and it gives me a major nerd-on.

Re:What I learned from the article

[Mozk]

Define "most"

Most of the applications that were benchmarked in the article.

Define "most"

Re:What I learned from the article

[itzdandy]

Though *MOST* of my tasks are not I/O bound, the ones that I wait on the most ARE I/O bound.

Seems that the controller here needs a clock speed boost. there is no reason this thing shouldnt be able to crack 550MB/s over 2 SATA300 interface other than a slow controller.

This review should have also included a software RAMDISK running off system memory. It would be interesting to see what the chipset's capabilities are vs this thing.

Re:What I learned from the article

[fast turtle]

how about for those pesky temp files that Windows seems to always want to keep around like the Page/Swap file and Temp Internet Files along with all the other /temp files. Instead of having those problems, simply place em into an 8GB ramdisk that looks like a proper drive to Windows and get rid of the battery.

Re:What I learned from the article

[Tycho]

Would heavy I/O tasks, include the heavy use of the pagefile in a bottom of the line Intel system without enough physical RAM, running XP or Vista and from just about any major computer manufacturer? I wonder how long an MLC flash drive would last and what the write performance would be? Note: this applies to both laptops and desktops and the power savings between a hard drive and an SSD are fairly minimal.

Sure, drives with SLC flash have features that mitigate some of the issues that MLC flash has, however, if you expect that SSD price per GB and total capacity to overtake hard drives anytime soon, you are in for a bit of a disappointment. Hard drive manufacturers are doing their best to increase storage on mechanical drives and hard drives have plenty of ways available to increase their storage capacity.

Re:What I learned from the article

[BikeHelmet]

Bingo.

That's why I'm holding out for a FusionIO. Their cards go through PCIe4x - not SATA. I want 600MB/sec reads/writes! 60k ops/sec, and cheaper than this thing!

I just hope their consumer grade products perform as well as their enterprise ones. Apparently tech report will be reviewing them as soon as they can get their hands on them.

Re:What I learned from the article

[linhares]

What use is it to boot from a ramdisk like that? The CPU cannot execute zeroes.

If Apple sells that, they will buy.

Re:What I learned from the article

[conureman]

I'm not yet comfortable with the long-term reliability of the current SSD units, vs: I haven't personally experienced DRAM failure in ten years of hardware use.
Certainly the remote NAS is a major part of the solution, as the DRAM partition is categorically unreliable.
I don't even know what is "bottle-necking" the process now, the rig he's running is a generic HP, every single bit of it makes for lack. It seemed to me, after I'd addressed the issues of going to 64 bits, and adding a few processor cores, and all the RAM that the budget and new mainboard would allow, &C., that maybe the best place to improve performance is storage I/O. This seemed like the least expensive way to speed it up, but I haven't yet done the math yet and it could need some "more better" engineering.

Re:What I learned from the article

[mcrbids]

That's not what I want. I don't want to have to buy the fastest, most expensive RAM available just to use as a RAM disk. I'd prefer cheaper RAM, maybe two or three generations old, that I can get in massive quantities.

Except that you can't get two or three gen old RAM in massive quantities. At least, not for as cheap as the new stuff. See for yourself: 1 GB of PC 133 RAM is more expensive than 1 GB of DDR or DDR2 RAM. There's a very short window of "cheaper" just behind the bleeding edge that's cheaper than the very latest (DDR3) but new motherboards support this type of RAM too, negating the "two or three gen old" situation that you state.

Most people think that the older the technology, the cheaper it gets. But this is only true for a very small time window, at which point the older technology gets phased out (not profitable, anymore) at which point it becomes a "niche" marketplace with very low volume and very high prices.

Example: A PC-133 RAM stick needed to keep a $12,000 vertical-market weaving loom operational, where the cost of the additional RAM pales compared to the cost of the entire integrated system. If you need that extra 512k of RAM in your $12,000 loom, paying $100 for it isn't such a bad deal.

But there aren't many people stuck with $12,000 looms, so the price of the older technology skyrockets until it's simply not available anymore. (Ever try buying a *NEW* Cx 6x86 processor in the last year or so?)

Re:What I learned from the article

[Ed Avis]

Just get rid of the swap file - it is not needed if you have enough physical RAM. And get rid of Temporary Internet Files by telling IE not to use a disk cache (again, with enough physical RAM it can cache in memory instead). In both cases adding this device doesn't give you anything you can't get by adding real memory.

As I mentioned, if you really want a RAM disk, you can get your OS to create one from RAM...

Re:What I learned from the article

[xtracto]

The main reason is the underlying technology used for the memory.

The 32GB SD card uses MLC flash technology, achieving higher memory storage in less space and at less price at the cost of performance while the 64GB SATA gard disk uses SLC technology which is more expensive (per MB) but also faster and more reliable.

See: http://en.wikipedia.org/wiki/Flash_memory

Re:What I learned from the article

[AmiMoJo]

A 4x PCIe card will not fit into a 1x PCIe slot.

Actually, most server grade or higher end enthusiast mobos have open ended PCI-e 1x slots, so 4x cards will fit. You can also just use a 16x slot, which even cheap mobos have.

As for your assertion that many mobos don't have PCI-e, the chances that this kind of drive would be used in a 1U server are slim. Data on the drive is not particularly safe compared to traditional SSDs or HDDs, and most server mobos have lots of RAM slots anyway. In most applications having lots of RAM is going to help more than having a SATA RAM drive. On-board RAM is much, much faster than this thing, and in fact so are SSD or HDD RAID arrays. Virtually all servers, 1U or otherwise, have a PCI-e slot these days for RAID cards anyway. Half hight or full form factor at 90 degrees.

All modern laptops have mini PCI-e slots, and have done for years. There is usually a wifi card installed in it. A lot of the higher end ones have multiple slots, for things like SSDs (used for ReadyBoost in Vista to supplement the HDD), GPRS modems and even graphics cards. RAM disks use much more power than SSDs. Multiple DRAM modules that require constant refreshing vs. SRAM/flash memory that requires nothing to maintain it's contents.

Re:What I learned from the article

[SanityInAnarchy]

It seemed to me, after I'd addressed the issues of going to 64 bits, and adding a few processor cores, and all the RAM that the budget and new mainboard would allow, &C., that maybe the best place to improve performance is storage I/O.

It seems to me, at that point, it's probably a software problem. Not that faster storage would hurt...

Re:What I learned from the article

[cstdenis]

> all OSes support the idea of a RAM disk...

Windows XP can't make one. And the only free third party program I found to do it under XP caused serious stability problems.

Just because *nix systems can do it well, doesn't mean all OSes can.

Re:What I learned from the article

[severoon]

Never used a Mac before...I'm scared.

Seriously, I'm honestly thinking about moving off of Windows completely and onto linux. The only thing I use Windows for is Photoshop work, and I have legitimate access to a legal copy (-gasp-), but only for Windows...this is out of my control. However, if I could get a butt-kicking machine like the one I describe above...linux + some kind of virtualization solution might just be the thing...

My experiences with Macs have been less than fruitful. I don't like the OS—it does too many magical things for my tastes that don't make any kind of sense. Maybe it's the result of growing up in a Windows world, but if I'm going to learn a new agglomeration of random decisions, I'd prefer it to be not just different than MS's, but somewhat more cohesive.

Re:What I learned from the article

[jsoderba]

With current RAM prices it is cheaper to correct the RAM deficiency in such a system than to add a SSD, unless you're hitting the 4GB limit and can't move to a 64-bit OS for some reason.

Re:What I learned from the article

[pclminion]

True, but if some kind of RAM-drive market takes off, all of a sudden it might still be profitable to continue producing RAM using technologies that are 12-18 months old... I won't hold my breath, though.

Re:What I learned from the article

[PipsqueakOnAP133]

I'm not sure if it's the ACard unit having a chipset that can't reach 3Gbit/sec or that the 3Gbit/sec spec of SATA includes overhead.

For example, USB2 never reaches half of 480mbit/sec in a benchmark because of overhead.
Firewire 400 comes close but I've never seen it get all 400mbit/sec.
1000baseT and 100baseT both haven't ever maxed for me on a benchmark either.

Re:What I learned from the article

[symbolset]

The real solution to the disk/Ram problem is to rethink how we use these devices.

Have you seen the specs on the new fusion-io gear? 320GB, enough bandwidth to saturate a PCIe X4 and 220,000 IOPS read and write. IBM recently RAID configured some to get 1M IOPS.

If you've got the cash, there's no longer any reason to be IO bound on the local disk. We don't have to make excuses anymore about how we don't write to local disk anyway. That's good because with cores per box stretching to 32 and doubling in the near future, the excuse was getting pretty thin.

Why are these always so expensive?

[rolfwind]

Without RAM, this costs $380 which is probably more than double the RAM itself if you don't use anything to extravagant. I know other companies offered these in the past, with the similiar high price, always to act as a harddrive with a battery for backup. It was always easy in linux to make a portion of your memory act as a ramdisk, however many motherboards often didnt enough ram slots to make it appealing to split memory up like that.

I wonder if a company like Apple can instead, on its laptops for instance, just move to SSD for its laptops since they are becoming seemingly cheap, exclusively (for OEMs) license a technology like MFT, and get a real speed edge on other makers. I think it would make more sense than a ramdisk where the bandwidth of ram vs the hard drive channel seems overkill.

Re:Why are these always so expensive?

[tulcod]

what bothers me most is that these RAM disks aren't even much faster than intel's SSD. considering the price, i think that's quite a shame. long live competition.

Re:Why are these always so expensive?

[Bearhouse]

Volume. Will always be a niche product, so they have to sell it at a high price. Now, if Dell or somebody did a buyback scheme of their old PCs and recycled the memory in some kind of cheaper version of this box...

Re:Why are these always so expensive?

[seven of five]

Yes, but SSDs are slow at random writes - to the point that it negates the benefits of random reads leaving you with a system about the same speed as harddrives.

Yeah, but it's nonvolatile.

oh, wait...

Re:Why are these always so expensive?

[Fweeky]

Intel's new SSD's have MFT-style stuff on the controllers; they have very fast random writes.

In contrast, the Intel SSD does about 8,500 4kB random writes per second.

Re:Why are these always so expensive?

[afidel]

Uh, the X25-e is about $700 and isn't nearly as fast. Of course as I found out today it's faster than the freaking $500 HP P400 raid controller in my server, plug the x25-e into the builtin SATA port on my HP desktop and I get ~40K IOPS and 170MB/s @4K 100% random writes, in the P400 I get ~16k IOPS and 65MB/s same workload.

Re:Why are these always so expensive?

[ZoCool]

What bothers *me is that no real progress seems to have happened since I upped my Mac 128 to a Mac 2048, and ran a 1Mb ram disk that held the OS, a ram-tuned WP (WriteNow, NeXT's default WP) Excel & room for one more. PC users used to laugh at me lugging my 9kg about - sooo slooow they laughed - but couldn't understand the concept of a ram disk. That baby *worked almost as fast as my current MacBook Pro! 25 years ago (as of yesterday *and I fired it up *and it still works.) (Ahem, except for the occasional 400k floppy access to save, of course.) I've dreamt of being finally able to get back to working in a ram disk for years, and it's almost affordable at last.

No ECC...

[KonoWatakushi]

so, this is just as worthless as Gigabyte's i-RAM.

Re:No ECC...

[cyberjock1980]

Actually, this box supports it's own ECC function. If you use ECC RAM, the ECC function is just like a normal ECC function. However, if you don't want to spend extravagant prices for ECC RAM, the box will create it's own ECC function. This function does use 1/9th the total RAM, but it is ECC, and it works. I own 3 of these boxes right now. They are fast as heck. I haven't played with SSD much(my first SSD drive arrives in the mail today), but I was able to perform tasks at performances that were beyond comprehension. From the time I double clicked the executable to the computer completely rebooted after the install was less than 2 minutes. Currently, Windows XP with all the standard software installed is less than 30 seconds.

Re:No ECC...

[Skapare]

Doing ECC the way this box would have to do when non-ECC RAM is installed, means that not only is it using 1/9 of the memory for ECC, but it also has to store it separately, requiring extra RAM access cycles, and potentially slowing down as a result.

The article says ECC DIMMs are NOT supported. Are you saying the article is wrong?

Re:No ECC...

[cyberjock1980]

The manual I got with mine says ECC RAM is supported. I can't confirm that ECC actually works as I have no ECC memory to test at the present time.

Re:No ECC...

[(H)elix1]

I've got one. Registered ECC is not supported. Unregistered ECC is supported. I saw no real performance decrease in simulated vs real ECC RAM. The SATA interface seemed to be a much greater bottle neck.

Re:No ECC...

[cyberjock1980]

Sorry, 12 hour work days fragmented my last post.

The time to install Windows XP SP3 from SP0 was 2 minutes. This is a significant difference from the typical 20-30 minutes(sometimes more) I've seen in the past for the SP3 installation. Everything is so much faster. My computer is a core i7 920, and when I use my platter drives to boot instead of my ANS-9010, it's night and day difference. The difference in performance to me is akin to telling your family member to upgrade from 512MB of RAM to 2GB. It's night and day. Plain and simple.

Re:No ECC...

[drsmithy]

This function does use 1/9th the total RAM, but it is ECC, and it works.

I'll take a stab in the dark and say their chipset is doing RAID3 or RAID5 behind the scenes to "emulate" ECC. So you're probably losing 1/8th (a single DIMM) not 1/9th.

Re:No ECC...

[cyberjock1980]

Nope. It's 1/9th. Read the manual on their website. They really use ECC. The size deteced in the BIOS is 1/9th less than expected.

Re:No ECC...

[fyonn]

I'd like to see someone make an expresscard (3/4) with as much ram as possible available over the PCI-e port as a disk. this could be used in laptops as swap space and temp space. I don't know if it would need any battery power, I know the machine could format it on boot, but what about when the thing sleeps? would there be enough power to keep it going?

anyways, if one could get 4 or 8G of ram it could be very useful. doesn't have to be extra fast for ram as it would be very fast for disk.

as long as it doesn't stick out of the slot so my MBP looks nice and sleek :)

Re:No ECC...

[bill_mcgonigle]

ZFS would help.

ahh 1993 was a good year

[bigattichouse]

1993 was fun. A few buddies and I built out a hardware ram disk with an unused 486 and some serial cables. Granted, it didn't come in as a native HDD, or even as a linux file system, but we built a little API that made it work reasonably well in software with standard file operations and even a crappy little "directory" structure. It worked fairly well... for some reason Mondo2000 pops into my mind.

Re:ahh 1993 was a good year

[setagllib]

These days you can just use iSCSI to any free Unix-like and export a memory-backed virtual disk. It's also a nice way to use one machine's memory as swapspace for another, and with a fast network link it's like having more RAM in the client machine.

Great for swap and /tmp

[kimvette]

This seems like it would be an excellent solution for a swap drive and for a c:\temp or /tmp directory.Fair enough as swap *nix OSes will reap little benefit, but Windows seems to hit swap no matter how much RAM you have so there should be some significant performance gains there.

Re:Great for swap and /tmp

[Ginger Unicorn]

The whole thing is pointless - why not just put 64GBs of ram in your PC and let it fill it up with disk cache. This makes no sense. If you compare this thing to just putting the RAM in your PC there are NO upsides. The data is vulnerable, it's massively expensive and an inefficient use of the RAM modules. Madness.

Re:Great for swap and /tmp

[Ginger Unicorn]

Also - if you had 64GBs in your PC there is plenty of space to create a RAM Disk in memory if you specifically needed one.

Re:Great for swap and /tmp

[Kjella]

I agree. There was a time when it made sense to use an i-RAM to get around the 4GB limitation, but with 64 bit readily available I wouldn't begin to consider this unless you need >16GB RAM. Even then you're likely to find server motherboards with more slots and support for higher densities, not to mention more sockets - I would imagine that that most applications which need this amount of RAM also could do with more processors.

Re:Great for swap and /tmp

[ustolemyname]

I wouldn't say "no" sense. It's battery backed up + connected to a compact flash slot, so when the power goes out it starts backing up your data to permanent storage.

My apologies - forgot I wasn't supposed to RTFA.

Re:Great for swap and /tmp

[cyberjock1980]

If you compare this thing to just putting the RAM in your PC there are NO upsides.

Ok...
1. Find me a motherboard that has 8 RAM slots that doesn't require expensive ECC and/or Registered memory
2. Find me a computer that can boot from it's own RAM drive.
3. Find me a computer that can use a RAM drive that can be persistent through reboots without having to save the contents to something else.

I have several of these, and I run a power cord that is normally used for one of those SATA/IDE to USB kits in the back of my computer to power my box.

You don't think about all of the uses that this thing offers.

Re:Great for swap and /tmp

[wed128]

But just look at those doom 3 level load times! I DON'T HAVE 7 SECONDS TO SPARE!

(also, i use solid gold SATA cables so my data doesn't get dirty)

Re:Great for swap and /tmp

[somenickname]

A big upside to this approach is that if you are actually able to saturate the SATA 2 buses with the drive, you are in luck because you are now using a bus that's a few orders of magnitude faster.

Re:Great for swap and /tmp

[Hinhule]

Because as he said windows seems to swap no matter what.

What you could do is use it as system memory as you suggest, set it up as a RAM drive and use that as your swap disk.

Re:Great for swap and /tmp

[dlinear]

I don't know about you but in my computer I only have 2 slots for RAM. This device makes a lot of sense to me as I have been pining for a DDR2 version of this device for quite some time. Mainly because DDR2 is dirt cheap compared to other memory modules.

Re:Great for swap and /tmp

[kabocox]

The whole thing is pointless - why not just put 64GBs of ram in your PC and let it fill it up with disk cache. This makes no sense. If you compare this thing to just putting the RAM in your PC there are NO upsides. The data is vulnerable, it's massively expensive and an inefficient use of the RAM modules. Madness.

Well, from everything on this product that slashdot has mentioned, just sticking RAM on a motherboard would be a better solution. It's not always the best though.

I've wanted one of these things for awhile:
http://www.devhardware.com/c/a/Storage-Devices/CENATEK-Rocket-Drive-SSD/
But could never justify the cost to myself. Were real RAM drives comes into their stride is any app that is HD I/O bound gets hugely speed up. You've also got to consider that this stuff is scaled down industrial stuff. I'd glance at the real 64 GB RAM disk and the all cost around 40-50K. They weren't for joe slashdot home user unless you had a few tens of K you wanted to spend. Now the real stuff had a builtin HD that the RAM was mirrored and it wasn't that difficult for them to convince business users to use a good UPS on the thing. You wouldn't believe the differences sticking any DB app on one of these things makes. Trust me you know when you really need one of these.

Now where Cenatek came along they tried to cut the cost for a PCI plugin board to $500-600 and then charged you a bit for the different amounts of RAM you'd put in their device. Sure it had to be externally powered or go dead. That's a draw back. But it did do "cheaply" what the real RAM drives did for the big boys. So if you really could afford it, you could stick your OS and favorite apps on there and notice a very responsive increase. They really started selling that thing somewhere between Win2000-WinXP when anything over 1 GB was rarely seen in a desktop. (It was easier adding a PCI card with 4 GB RAM than changing mother boards.)

Now a days with 3-4 GBs in "budget" desktops, I'd want a 64 or 128 GB RAM drive, but I'm also kinda like you, if I had the money I'd most likely see more immediate bang for the buck just adding system RAM. You do see really big increases though in real RAM drives. Then again how many mother boards do you see that'll let you plug in 64 GB of RAM? Actually, I think that some of this is just a stop gap especially at any individually affordable price. Just wait until you see 128 GB RAM in the $200 walmart special desktop.

Re:Great for swap and /tmp

[oldspewey]

2. Find me a computer that can boot from it's own RAM drive.

A 4-hour battery life on the ANS-9010 essentially destroys this argument. The remaining two arguments are valid however.

Re:Great for swap and /tmp

[hesiod]

With a 32-bit processor you have a 4GB RAM limit.

Re:Great for swap and /tmp

[(H)elix1]

Wish I could.... Many of the current motherboards have an 8G max, so 4x2G is all the RAM you can stuff into them. For those who don't have the system RAM to spare for a ram disk, it is another option.

For those who do, I've had good luck with this.

Re:Great for swap and /tmp

[Hal_Porter]

[Walks away from well.]

Madness.

Fixed that for you.

Re:Great for swap and /tmp

[Hal_Porter]

You know the monsters have to wait too, and it annoys they just as much.

Re:Great for swap and /tmp

[Glonoinha]

ramdisk.tk - I've been using it for a while and it works incredibly well. It has a freeware trial version (nagware that's not overly intrusive) so you can play with it to see if it fits your needs.

I use it for my browser cache in Firefox and as a scratch area for editing pictures (basically temporary working space for data I would discard when I was done with it anyways - so no worries about data loss in the event of a system outage.)

Re:Great for swap and /tmp

[sshir]

Adding to another reply.

1. Find me a motherboard that has 8 RAM slots that doesn't require expensive ECC and/or Registered memory

You don't need 8 slots. Just use metaram.

3. Find me a computer that can use a RAM drive that can be persistent through reboots without having to save the contents to something else.

That's the thing - you don't need to. Your application should be able to use it naturally. Just make sure you have a modern CPU which can handle it without performance degradation due to paging complications.

Re:Great for swap and /tmp

[ustolemyname]

Ostensibly, that battery is only there to give the device power while it backs up your data to a compact flash card.

Of course, you already knew that cause you read TFA, right? right?

Re:Great for swap and /tmp

[The_Wilschon]

So section off a bit of your memory as a ramdisk, and put the swap file there. Still going to be better than the SATA bottleneck.

Re:Great for swap and /tmp

[oldspewey]

Actually I did RTFA (or at least I read 3/4oTFA), and it left me rather underwhelmed.

Re:Great for swap and /tmp

[ustolemyname]

What, the fact that the ancient Pentium 4 was cause I/O bottlenecks? Agreed, their test hardware was very underwhelming.

Re:Great for swap and /tmp

[ajlitt]

IA32 can address up to 64GB with PAE.

Re:Great for swap and /tmp

[drsmithy]

The whole thing is pointless - why not just put 64GBs of ram in your PC and let it fill it up with disk cache. This makes no sense. If you compare this thing to just putting the RAM in your PC there are NO upsides. The data is vulnerable, it's massively expensive and an inefficient use of the RAM modules.

The data on this device will survive an OS crash or reboot. The data in your PC's RAM will not.

Madness.

No, niche.

Re:Great for swap and /tmp

[pipatron]

You could even use it back with kickstart 1.3. And the Amiga is a modern computer, just that it was launched 20 years before everything else. :)

Re:Great for swap and /tmp

[Belial6]

3. Find me a computer that can use a RAM drive that can be persistent through reboots without having to save the contents to something else.

That would be any computer running Amiga OS4.

Re:Great for swap and /tmp

[Lennie]

or you could just put /tmp on a ramdisk (the OS kind, not the 'hardware' kind).

Re:Great for swap and /tmp

[Ginger Unicorn]

That isn't an advantage - it's more of an overly elaborate kludge to ameliorate a disadvantage.

Re:Great for swap and /tmp

[Ginger Unicorn]

And what advantages does you RAM drive provide over simply having a very large disk cache combined with regular hard drive?

Re:Great for swap and /tmp

[zippthorne]

Please point us to a reasonable costing motherboard that supports 64 GB of RAM.

Seriously. That sounds pretty awesome to me.

Re:Great for swap and /tmp

[AK Marc]

The whole thing is pointless - why not just put 64GBs of ram in your PC and let it fill it up with disk cache.

Getting 64 Gbps of system RAM is not easy for a consumer PC. Also, disk cache is something that is often horribly used. It is handled by the OS, and the OS makes guesses. If I *know* there are some files I would use repeatedly and the OS won't let me mark them "cache preferred" (I've never heard of that being an option, but it if was, then you'd have a stronger case), then they may get pushed out of cache for stupid stuff like watching a movie stored on disk and having that cached (like you are going to watch it several times in a row or skip around in it repeatedly).

If you compare this thing to just putting the RAM in your PC there are NO upsides.

OK, so tell me what PC that is consumer oriented that will take 64 GB of ram.

Re:Great for swap and /tmp

[swilver]

Yeah, Amiga was able to do that since the early 90's, I remember quite well booting from a RAM disk to use newer Kickstart versions :)

Re:Great for swap and /tmp

[this great guy]

1. Find me a motherboard that has 8 RAM slots that doesn't require expensive ECC and/or Registered memory
2. Find me a computer that can boot from it's own RAM drive.
3. Find me a computer that can use a RAM drive that can be persistent through reboots without having to save the contents to something else.

1. Registered DDR2 is much cheaper than this hardware RAM drive based on unbuffered DDR2. Compare 8 2GB modules of reg DDR2 (16GB * $15/GB = $240) vs. this RAM drive + 8 2GB modules of unbuf DDR2 ($379 + 16GB * $10/GB = $539). And for 4GB modules, compare 32GB of reg DDR2 (32GB * $25/GB = $800) vs. RAM drive + 32GB of unbuf DDR2 ($379 + 32GB * $25/GB = $1179). And good luck finding 4GB unbuf modules, they are much less common than 4GB reg DDR2 modules !
2. Buying a RAM drive only to improve boot times is ridiculous. Something a teenager nerd would do to impress his friends. People who actually need (not want) data stored in RAM (whether it is a RAM drive or system RAM) usually do it for, you know, improve the perf of /O intensive applications.
3. As explained by the GP, this is already how the buffercache works on Linux: the kernel uses (very efficiently) system RAM to cache filesystem data, and will automatically cache more and more data as the system is used after a cold boot. Maybe you just don't realize how efficient this is. When launching OpenOffice on my Linux workstation, the first time is really slow (cold cache), but the 2nd, 3rd, etc, times, it hardly generates any disk I/O activity.

Not to mention that hardware RAM drives have laughable throughput speeds: 400 MB/s for the one reviewed in TFA, compared to 12800 MB/s or 10667 MB/s for standard system DDR2-800 or DDR2-667. I totally agree with the GP, what's the point of putting memory modules behind a slow I/O interface, with all the SATA protocol overhead ? What's the point of absolutely trying to prevent the OS from using these RAM modules as standard system RAM ? Just put the damn things in the motherboard's memory slots and let my OS use them as I want: system RAM, buffercache, software RAM drive, etc.

Re:Great for swap and /tmp

[cyberjock1980]

Your calculations compare the cost for 16GB of DDR2 Registered RAM to the ANS-9010 + RAM cost. You should actually be comparing the ANS-9010+RAM to the cost of everything necessary to make a RAMdrive using a system. So you will need: very expensive motherboard with 8 RAM slots, an expensive processor, hard drive, CD-ROM, power supply, case, operating system, 16GB of RAM, the electrical usage, the cooling needed, etc.

You'll see that the ANS-9010 with 8*4GB sticks is probably cheaper than running a whole second computer with a xeon processor and a single(or likely dual) processor motherboard. Plus the invested time in setting up the darn thing. But even after that, the performance of the ANS-9010 would more than saturate gigabit LAN. A system RAM ramdrive would be limited even by 10 gigabit LAN. So how do you intend to use this 'ultra fast ramdrive' in another system of your choosing if you can't really use the speeds to begin with?

I would agree that if you were building a computer it might be cheaper to have a bunch of ram used as a ramdrive and shared, but you still couldn't boot from it as well as save the data at shutdown without having to back up the RAMdrive during shutdown.

Re:Great for swap and /tmp

[this great guy]

You haven't checked hardware prices in a while, have you ? A "very expensive" motherboard supporting 8, no wait... 16 RAM slots costs $80 ! Add a single $180 socket F CPU to make use of 8 of these 16 slots, and this is still way below the price of the ANS-9010. Want to add 8 more RAM modules ? Another ANS-9010 will cost $380, or you can pop in a 2nd CPU in that mobo for $180. When I say RAM drives are overpriced, I mean it !

Regarding your comments that "400 MB/s is enough", you are wrong as you are completely ignoring many workloads where the throughput is needed internally to the system. For example, when processing data on the RAM drive itself, when rsyncing files with a remote machine where you need fast internal I/O throughput to compute the MD4 checksum, when searching through GB of data, when re-indexing a database, etc.

Re:Great for swap and /tmp

[this great guy]

Also, concocting a scenario where a 2nd machine would be restricted to act as the "RAM drive" is a laughable unfair comparison. Obviously combining the 2 machines in 1 is the way to do it.

Re:Great for swap and /tmp

[cyberjock1980]

Did you notice that there was a total of zero people that gave that board a 5 egg rating? And all 3 comments had less than spectacular things to say when it came to using that board.

I will agree with you that it's overpriced. Price is a bit steep. I'm just not sold that I'd really intend to build a computer using that motherboard. Everyone's comments on newegg had much larger cons sections that positives. Add that with not one 5 egg rating(yes, I see there's only 3 ratings), but I'm not sure it would be an investment(gamble?) I'd be willing to take. Can you honestly tell me you'd buy a motherboard that people claim is picky about which RAM it wants to use, isn't compatible with 1 person's raid controller(out of 3 comments) and your CPUs might not be supported because of a stepping issue. If I built a computer with that motherboard I'd be disgusted to play games on it, or anything else for that matter. PCI-X slots, you got to be kidding. 1 PCIe 16x slot(is it PCIe 1.0 or 2.0?). Where's the other PCIe slots I would need? That motherboard is quite outdated, hence the cheap price.

I'm not an AMD fan, but I'm not an Intel fanboy either. Me personally, I'd never buy AMD. But with how many friends have them and enjoy them, I don't find it justifiable to bash AMD either. When I looked at a $100 8 RAM slot motherboard, I found more people giving the motherboard 1 egg compared to 5. That's not such a great thing IMO.

I'll stick with my ANS-9010 and you can stick with your setup. I can use my ANS-9010 with any board that supports SATA, and your ideas are limited to buying into a build that limits your future choices(and perhaps compatibility with other cards). There's no effective way to boot from it, save it's contents in a reasonable period of time, etc. I feel my idea is more versatile and obviously much less committed to a particular motherboard/CPU.

From my personal experience, booting from the ANS-9010 and running my system off of it is significantly faster than anything I've ever seen. I can't even believe it's the same computer when I image my ANS-9010 to my actual hard drive and boot from it. Even after defragmenting and everything else, it just doesn't feel like the same computer. Having microsecond seek times to read data is significantly more important than the quantity of data you can transfer.

In summary

[Max Romantschuk]

Skimming the article, I'd summarize as follows:

Real world performance not radically better than fast traditional HDs or SSD solutions, and you can't power off your PC for the night. (Unless you backup to flash every night.)

I'd say this is a niche product, but could be a very good one for a chosen few applications.

Re:In summary

[doti]

what's the difference from a tmpfs then?

Re:In summary

[Max Romantschuk]

what's the difference from a tmpfs then?

There's a battery that lasts a few hours or something, but not through the night. RTFA for proper details. ;)

Re:In summary

[gmuslera]

They have a battery (lasting around 4hs) so survives turning off the pc, or do some not very extensive hardware maintenance. And maybe more important, rebooting (kernel upgrades happens).

In the other hand, tmpfs have the transfer rate limit of ram (~6G/s), while in this devices the limit is the SATA one (~400M/s).

Re:In summary

[Agripa]

and you can't power off your PC for the night.

I was a little surprised they did not include a way to tap the ATX 5 volt standby voltage.

Re:In summary

[elashish14]

If you're using tmpfs on a 32-bit machine, you're limited to 4GB or whatever the limit is on installable RAM. RAM disk eliminates this limit using SATA ports, but limits it to the speed of SATA transfer too.

I RTFA and now speculate

[moteyalpha]

It seems that with a little firmware it could be coaxed to do some content addressability. Considering that it is 10x faster inside than the peak of the SATA interface. It seems to me that there is a lot of potential. I always liked the ram disks when they were popular ISA cards and this could be the thing that could use the full power of USB 4.0 [sic]. Applications could be changed to take advantage of this speed. If lists and SQL databases could be sorted on the drive without CPU overhead, it could be very useful.

Re:I RTFA and now speculate

[morgan_greywolf]

Applications could be changed to take advantage of this speed. If lists and SQL databases could be sorted on the drive without CPU overhead, it could be very useful.

Until the power suddenly goes out and the UPS fails...ouch!

Re:I RTFA and now speculate

[morgan_greywolf]

And before any pedants out there note that I didn't RTFA, I did: 23 minutes for a CF backup is just not fast enough to cover a failed UPS.

Re:I RTFA and now speculate

[fuzzyfuzzyfungus]

It is when the device in question has 4 hours of battery backup onboard. Not something that I would want to have to get up in the middle of the night to deal with; but perfectly doable. Ideally, they could get this thing's power draw low enough to feed off the +5volt always-on rail; but battery backup is a start.

Re:I RTFA and now speculate

[Bearhouse]

Until the power suddenly goes out and the UPS fails...ouch!

Indeed, but then again if you have this kind of double failure, then you'll probably be worried about other things ;-) The drive has it's own on-board battery backup BTW...

Re:I RTFA and now speculate

[Bearhouse]

Agreed. Then again, this is clearly not designed for any kind of mission-critical / server application. Who wants to have to remember to insert (the right size) flash card in the thing, and press a button, when all the servers are going down, the UPSs are silent and the users are screaming?

Also, as noted above, the device would be useful for such applications only if it had more onboard inteligence and could thus unload he main processor....

So, we're agreed that it's a cool Geek toy then? Nice though...

p.s. Where would /. be without the pedants?

Re:I RTFA and now speculate

[mi11house]

This box definitely needs some firmware/FPGA tuning if it's not able to saturate its SATA connections.

Add to that the fact that the two-channel "RAID" configuration is *nowhere near* twice as quick as the single bus - this thing urgently needs some performance tweaks to fully realise the potential of the concept.

Re:I RTFA and now speculate

[poot_rootbeer]

Considering that it is 10x faster inside than the peak of the SATA interface. It seems to me that there is a lot of potential.

Since the performance bottleneck here seems to be the SATA interface, and not the memory itself, maybe the next evolution of this design could move the RAM directly onto the motherboard, and the CPU could be given rapid access to it through the northbridge?

Imagine what kind of performance that (purely theoretical for now) configuration could accomplish!

 

Re:I RTFA and now speculate

[(H)elix1]

I've got one. In practice, it is. You cut power to the device, it will start making a backup using its internal battery - which lasts 3-4 hours. This is not dependent on you pushing the button to do a manual backup of the current drive image.

Re:I RTFA and now speculate

[morgan_greywolf]

Mod parent +5, informative! Ah! That's what I was looking for! I didn't see that in the article. Thanks for that.

Re:I RTFA and now speculate

[Lennie]

Why do you need to press that button ? If the (outside) power is cut to the device, it will start it's backup using it's internal battery.

Re:I RTFA and now speculate

[Bearhouse]

Must have misread the article - did not see that it was automatic in case of power failure?

Being more practical is a bad thing???

[belligerent0001]

"However, it also highlights the device's shortcomings, including the fact that SSDs are more practical for most applications."

I'm sure that there is some sort of typo somewhere. Since when is being "...more practical for most application..." considered a shortcoming?

Re:Being more practical is a bad thing???

[LMacG]

The "ANS-9010 RAM disk" is the device with the shortcomings. SSDs are (in the author's opinion) more practical.

Re:Being more practical is a bad thing???

[bogaboga]

...Since when is being "...more practical for most application..." considered a shortcoming?

When any attempt to making use of the results of being practical is more of an exercise in frustration than maintaining the status quo.

Re:Being more practical is a bad thing???

[Anonymous Coward]

Since when is being "...more practical for most application..." considered a shortcoming?

I think the artical meant the FlashDrive bases SSDs, not the memory bases SSDs. Why the artical uses the term SSD, which points to both types of disks I do not know.

Re:Being more practical is a bad thing???

[belligerent0001]

If the implementation is an exercise in frustration, would that not, therefore, make it impractical to begin with?

Re:Being more practical is a bad thing???

[AK Marc]

A RAM disk is a drive, and it is solid state. So it isn't incorrect to call it a Solid State Drive. Now, if you are talking persistent drives, you'd have a different issue, but I don't see why calling any drive that's solid state an SSD.

Great untill a stick dies!

[xxMSAxx]

woa is the day you have a 64gig hdd made up of ram sticks and one goes bad....happy hunting grasshopper.

Pagefiles, watch out

[Twinbee]

Yay, a RAM HD! I'd like to see the pagefile dig into this - Microsoft must be foaming at the mouth. Sorry if that seems like trolling, but I've had it up to here with the constant and painful HD thrashing that Windows always seems to enjoy doing (and probably their less than perfect implementation of it).

Re:Pagefiles, watch out

[Ihlosi]

Yay, a RAM HD! I'd like to see the pagefile dig into this - Microsoft must be foaming at the mouth. Sorry if that seems like trolling, but I've had it up to here with the constant and painful HD thrashing that Windows always seems to enjoy doing (and probably their less than perfect implementation of it).

You'll be better off sticking all that RAM in your computer than in a $380 box. (Or, optionally, spend the $380 to upgrade your mainboard to one that can hold all the RAM).

Re:Pagefiles, watch out

[Twinbee]

Perhaps, but why do I have my suspicions that XP will even bother to use that memory, especially for the important stuff (like keeping standard requestors in RAM).

Re:Pagefiles, watch out

[Zironic]

Hmm, it would be really interesting to just put 4 more GB of RAM into the computer and make a ramdisk out of it and then point the windows pagefile to the ramdisk. That should hopefully once and for all put an end to that painfull trashing.

Re:Pagefiles, watch out

[Perf]

HDD trashing or thrashing?

But you did make a good point.

Not that impressive

[__aardcx5948]

In the real world benchmark, the only place where it shines is during file copy operations. It's on par with Intel's SSD when measuring OS load time, game level load times and other application benchmarks. The battery only hold 16GB for 4 hours too, you'll need that CF card to backup to (which is a neat feature, just the press of a button to backup and restore). Any way, I'd gladly take it for free.

Re:Not that impressive

[Glonoinha]

Actually in the real world benchmarks the only place that it really shined was in transactional processing of multiple client streams. In single threaded, single activity processing it isn't really that much faster than even regular hard drives. Save 10 seconds copying a few gigs of files or making a tar file from your existing data - what good is that? Shave 12 seconds from your boot time - again, what good is that? Now enabling your transaction processing system to handle 10x as many transactions in a given time frame with a simple hardware bolt-on - THAT is significant. Having your back end able to serve up 10x the number of end users at the same time - that's MONSTER gains.

Given that your application servers are on UPS infrastructure and you rarely boot them at all, much less leave them off overnight - this thing is a god-send to transactional processing systems in large-scale environments. For your average home user (or even 'leet home user) - maybe / maybe not.

A few years ago 64G of SSD would have run you way more than $100,000. At $400 delivered (plus memory costs) this thing is practically free. I'm considering getting one, and I'm about as cheap as they come.

Re:Not that impressive

[Wonko the Sane]

If you are running these type of applications, then why wouldn't you use a server that lets you install 64G of ram in the first place?

Is there any case where a 64G SATA ramdrive performs better than 64G of main memory?

Unless your OS really sucks at disk caching.

Re:Not that impressive

[Glonoinha]

Is there any case where a 64G SATA ramdrive performs better than 64G of main memory?

Sure - unplug the computer for half an hour.
Actually, simply power cycle the box (say ... after installing a service pack on Windows, or once each year in Unix just so you don't forget what the boot sequence looks like.)
Or - on a machine that's already running a ton of processes and using 32G of memory in the first place, adding an additional 64G of memory means having a machine that can hold 96G of memory - I've got some nifty toys, but nothing I have right now will let me stuff 96G onto the motherboard. This $400 device gives me almost the same thing, has a battery backup and the ability to back up to CF, and I can drop it in a machine running XP Pro or Windows 2000 if I want.

Don't get me wrong - I run a ramdrive (ramdrive.tk) at home and I love it. I'm a big proponent of ramdrives. I've been using them to optimize systems since the first time I loaded COMMAND.COM onto one on a dual floppy 386sx machine with no hard drive, circa DOS 3.31. It's just that all my current machines hold a max of 4G of physical memory (four slots - although I haven't actually tried to stuff 2G chips in one, so I might be able to go to 8G if the BIOS doesn't choke on it.) So I can consider dropping the memory in this thing and be able to move it from machine to machine if I want, or upgrade my personal hardware inventory to all new computers ... and as much as I want new machines, they are not in my budget this quarter (and next quarter isn't looking good fiscally either.)

I really wanted to get an iRAM, but the 4 Gig limit sort of dissuaded me (plus DDR is about twice as expensive as DDR2 right now.) If you could put 2G chips in the iRAM for a total of 8G per iRAM, I'd totally consider getting a few of those instead - but this SATA box, for the cost of about two iRAMs, will let me fill it with twice as much memory for about the same price. I'm considering getting one, to tell the truth.

What's the use?

[Woek]

Isn't it better (easier, faster) to just put 64 GB of actual RAM into your system (and if needed use part of it as a ramdisk), and using a normal drive for storage? You could have an enormous amount of disk cache!

Re:solid state

[Anonymous Coward]

Would you rethink your definition of fail if they were using a Sony laptop battery?

Don't see the point

[samael]

I can't see why I'd use this rather than just putting more memory in my machine...

Re:Don't see the point

[samael]

If you're the kind of person who's springing for this kind of thing then you're probably running a 64-bit OS. That'd be my first port of call in this situation anyway, unless there was a good reason for restricting myself to the 32-bit version.

Nothing new

[Burdell]

I have a high-load mail server that uses a 2G RAM disk (a Curtis Nitro!Xe) for the queue. It looks like a normal 3.5"/1" high SCSI drive with a SCA hot-swap connector. It was made before high-density CF cards, so it has a 2.5" notebook hard drive inside for storage after shutdown (it has a battery to start the drive, dump the RAM, and shut down). We've had this in service for almost 5 years, and it has really made a difference.

The point to a RAM disk is not necessarily bulk data throughput, but I/O operations per second. Mechanical drives are limited to 100-200 random IOPS or less, while the RAM disk can easily hit 100,000.

Re:solid state

[wisty]

Any drive can fail, that's why you do backups. Maybe put them in RAID-0 and pray the batteries don't all go :). At least rebuilding would be quick.

Good idea

[ledow]

My university used RAM disks back in the day - it was the only way to get decent performance on older machines. The computers didn't even have hard disks in. My brother (who went to the same university) has a story where he sped up his large FORTRAN compiles by a factor of 10 just by working out how to use the RAMDisk (which was only ever used by the PXE-style boot procedure and then hidden from the OS) for his own purposes and people couldn't work out how he was doing it because he still took stuff home and brought it in on floppy. This is a nice hark back to those times.

The killer, however, is the price... the price of a PC, basically, before you add the RAM. If you're REALLY serious, you'll have machines that can just take the extra RAM directly and do this in software. If you're not willing to pay that much, well, nothing will work for you but a few bits of extra RAM and a fast SSD for the same price won't go amiss. However, if you occupy the middle ground... this still doesn't seem worth the effort. It'd be cheaper to just buy an SSD, some extra RAM for cache and maybe even a cheap PC to throw it all in (if NanoITX supported 8Gig chips, this device could almost be made obsolete overnight).

The interconnect too - yes, it emulates a SATA drive but it emulates two as well and fails to do anything significant with them. So you'd need a RAID0 setup, with independent SATA setup, and an expensive device, with lots of even more expensive RAM just to be a fraction of a second quicker than an off-the-shelf SSD in the same machine. The people for whom it's worth it won't want to be bothered with all this.

The CF Backup feature is fantastic. I love the idea. But 20 minutes is a long time to wait if the battery is only four hours worth when it's brand new (four hours? At least 24 would have been useful and given you a chance to actually do something with it). You would want to be backing up anything this thing held anyway, so you don't really gain anything because the CF is the most inconvenient backup because of its manual nature.

I can't see a situation where 64Gb of fast storage is worth that amount of money + time + hassle + 64Gb of RAM + potential firmware problems + interface cabling + ... The bottlenecks in anything serious are going to be elsewhere.

Re:Good idea

[merreborn]

I can't see a situation where 64Gb of fast storage is worth that amount of money + time + hassle + 64Gb of RAM + potential firmware problems + interface cabling + ... The bottlenecks in anything serious are going to be elsewhere.

If you think this is pricey, you should see what the guys who run EVE online paid for the RAMSAN units they have backing their databases -- over $150 per gig. CCP claims a 4000% percent performance increase, as a result of the upgrade, however.

There are definitely plenty of IO-bottlenecked servers out there that could benefit dramatically from a good SSD solution. But yeah, if you're just gaming and posting to slashdot, a $600 consumer-grade SSD isn't really going to make much of a difference in your desktop rig.

Re:Good idea

[bennomatic]

I remember back at my old job at an ISP, we had some old Apollo boxes sorting our USENET feed, packing up subscriptions for our individual and UUCP clients. We got to a point where we were doing almost 30 GB of USENET data a day (full feed, all groups), and the swapping alone was effectively killing our servers.

We got a 100 MB RAM disk for the primary processor--cost a butt-load, IIRC--but using that as the swap volume, we were able to complete everything in something like 25% of the time.

Much as I like this, I can't see the market

[myxiplx]

First of all, I absolutely love these devices. It's a great idea that's been well executed, and yes, they're a niche product, but we've one or two apps that would notice the increase in speed from these, and if I had the money I'd buy a whole bunch of them to stick in our servers. ... except that you don't get 5.25" bays in an awful lot of modern rack mounted servers. Certainly none of our new kit has them; all that space is taken up with hot swap 3.5" or 2.5" drives.

And that's what kills it for me. Even if I'm looking at a new server I'd have to make sacrifices to fit one of these. My first choice for a new storage server is going to be one with 24x 2.5" drive bays. I'd have to sacrifice a full 8 drive bays to make room for one of these, and it's just not worth it. Not when I can buy an Intel SSD for the same price, loose just one bay, and have it hot swap to boot.

And even worse, there are PCIe devices just around the corner, with 3-4x the read and write speed of any SATA device. Those will drop straight into any of our current servers, no problem at all.

So unfortunately, much as I love the ANS-9010, I just can't see any reason to buy one :(

failure mode

[Lord Ender]

How does this thing handle getting the power cord yanked in the middle of a large write operation?

Re:failure mode

[(H)elix1]

Same as any HDD - a hard shutdown. The battery pack will then start backing up the current state of the memory to a CF card, so that when power is returned to the system you can run fsck or chkdsk. If you don't have a CF card, it will keep the RAM alive for a few hours, then all is gone if power was not restored.

Re:failure mode

[MightyYar]

How does this thing handle getting the power cord yanked in the middle of a large write operation?

Since you didn't RTFA, you'll be happy to know that you can keep on gloating. The thing has absolutely no backup mechanism, no battery, no ability to write to a CF card. If the power so much as winks, all of your data is garbage.

Re:failure mode

[suggsjc]

It woul

Re:failure mode

[MiniMike]

In the article it mentions that it has a battery backup which lasts for four hours. It would have to be a very large write operation to exceed that time. Other than that, it probably handles it as ungracefully as a regular hard drive would, except that it would still be able to save whatever was cached but not written (whatever that means for this drive).

Re:failure mode

[Kingrames]

It pulls out a gun and shoots the user.

For some tasks its very useful

[MindKata]

"I'd bother, because most of *my* tasks are disk I/O bound"

I totally agree. For some tasks its very useful. (If we had memory that was the speed and no write deterioration like it was DRAM, but non-volatile like Flash, then that kind of memory would make this DDR2 based product obsolete, but until that time, its very interesting. Maybe there's hope in time for Memristor based memory, but for now this product does have some advantages).

I was interested in Gigabyte i-RAM but its too limited at only 4Gb.

I have wondered how practical it would be to create an open hardware project like this product, based around FPGA and DDR2 (or maybe DDR3 as that will likely be cheaper within the next I guess about 18 months and taking over from DDR2). Does anyone know of any existing open projects like this?

Re:Pointless curiosity

[mariushm]

Compare 350$ (device) + 300$ for memory for a 32GB drive versus 4 x 600$ Intel SSDs.

Son of iRAM

[(H)elix1]

I got one of these in our lab, and can answer questions on it. Had both units... the 6 slot version and the 8 slot version. This thing is the spiritual successor of gigabytes's iRAM. It takes bog standard DDR2 RAM as storage and lets you connect it as a SATA drive.

A few of the things it improved on the old iRAM.

*DDR2 supported ram, with 6-8 slots, taking up to 4G sticks.
*A fair sized battery.
*A CF backup slot.
*RAID friendly, multiple SATA ports on 8 slot model.
*Uses 5.25" bay rather than PCI slot.
*ECC

First off, no special device driver was needed - the drive was OS agnostic. Every mainboard and controller card I used saw it the device like any other SATA hard drive you might plug in.

The RAM slots take bog standard DDR2 RAM. The documenation mentions speeds of 400/533/667/800 are all supported. Benchmarks with 533 and 800 grade RAM produced identical benchmarks, so faster RAM does not appear to have any impact. I also mixed and matched faster and slower DDR2 modules without issue.

Just like most mainboards, the RAM needed to be installed in pairs if over one stick was used.

Unbuffered ECC or non-ECC modules are both supported. Registered RAM was not. I tried to pull eight 4GB sticks from one of my Sun boxes to give the 'full montey' test. No joy. Had to stick with the far cheaper RAM.

There was an interesting option for these who wanted to have ECC but used 'regular' non-ECC RAM. Eleven percent of the memory could be reserved for error correction. Again, all hardware based - just move a jumper. Performance metrics between ECC and 'simulated ECC' had negligible differences.

The 8 slot model has two SATA ports. By setting a jumper, you could have the entire RAM capacity as one large drive on one SATA port or split it as two independent drives. If you splid the drive you had to have an even number of RAM sticks installed. Another jumper would dumb the interface down to SATA1 speeds rather than SATA2. Never tested that....

Did test RAID-0, however. (grin) The synthetic benchmarks don't hit this device's sweet spot - database usage. Reads are fast. Writes are just about as fast. The RAID controller really makes a difference, as my 3Ware card performed significantly faster than with the mainboard based RAID. Using a EVGA 780i mainboard, it was not crushingly faster than a trio of velociraptors.

For anyone who has installed XP, you know the wait between hitting the 'workgroup' and the first reboot? Just over two minutes. By far the fastest install I've ever done. The OS also started faster than any other disk or SSD system I've used.

The CF bay was a nifty option. The question came up - what if I want to shut my machine down overnight? You can. If you have a CF card with more capacity than your RAM, it will back up the disk image automagically. You can also push a button to back up the current 'drive image' to CF, and another to restore the image. (I was able to go back and forth from Linux and Windows very easily).

Anyhow, tis a fantastic high speed scratch disk or OS disk when write speed matters. For those of us who already maxed out RAM, this covers the gap between RAM drive sharing RAM with the mainboard and fast disk.

Re:Son of iRAM

[GameMaster]

Hrm, you sure you were using the same model as the one in the article? The one in the article had 8 slots that could handle dimms up to 8GB each and didn't support ECC ram.

Re:Son of iRAM

[drsmithy]

*Uses 5.25" bay rather than PCI slot.

This swings both ways. If what you have is a typical rackmount server, then a device that goes into a PCIe slot is far more preferable.

Realistically, they should be able to make two devices - PCIe-mounted and drivebay mounted - using practically identical circuit boards. Ideally, the PCIe device would have a disk controller 'onboard' rather than plugging into the motherboard's SATA ports (and therefore have better performance).

To be honest, I'm a little unsure of the target market with this product. The form factor and relatively slow onboard chipset mean it's not really aimed at the enterprise market, but the cost and massive IOPS performance don't really suggest it's a good fit for the low-end market either. I can only assume they're trying to make money out of the same people who think they need a triple-SLI of dual-processor video cards.

Re:Son of iRAM

[(H)elix1]

Got a ANS-9010 and played with a ANS-9010B. Registered ECC RAM does not work. Non-registered ECC RAM did.

Re:Son of iRAM

[Janek Kozicki]

Using a EVGA 780i mainboard, it was not crushingly faster than a trio of velociraptors.

Practical question - if you buy this ANS-9010 and 8 cheapest 4GB ram sticks - is it cheaper than three velociraptors?

Re:Son of iRAM

[afidel]

Dude, velociraptors aren't anywhere NEAR this, you'd need about 100 to get the same kind of I/O's per second. I know because one of my disk groups in my SAN is 110x 146GB 10k rpm drives and it gets about the same level as those shown in the benchmarks. Of course it also provides 8TB of RAID10 space which is fully redundant (I can lose an entire enclose and not lose any data due to the way the data is striped).

WAY over priced.

[michrech]

Get the device down to about $50 and I'll think about buying one. I don't see any reason a device like this is $380. That's just crazy.

Re:WAY over priced.

[rednuhter]

$380 excludes the ram OUCH!
(read the conclusions page)

Re:WAY over priced.

[michrech]

Errr.. I already did. That's why I suggested putting the price to $50 before I'd even think about buying one. ;)

$380 excludes the ram OUCH!
(read the conclusions page)

Re:solid state

[oldspewey]

Maybe if the battery life were longer than 4 hours, I'd be willing to overlook the fact this thing is based on volatile memory. But if I can't even shut down the machine overnight (and it takes ~15 minutes to restore 32GB from flash) then AFAIC this "SSD" is strictly a novelty.

Re:No ECC. & slower then a motherboard with 64

[hesiod]

With this, we'd also be lucky enough to lose our data too

That's why it has a CF slot. Leave a CF card in the slot, and if the shit hits the fan, go press the "Backup" button, which will take less time than your UPS' power.

BattDisk

[chill]

Ah, Amiga old buddy, I miss you.

This sound's like DKB's Battdisk. Next to a Video Toaster and SuperGen, the Amiga 2000s best friend. It took a machine from the slowest booting system in the age to the fastest.

Performance issues...

[Firethorn]

That is pretty mediocre.

For DDR2, maybe, for fixed disk usage? Still pretty good compared to spinning disks or even flash.

Of course, RAM disks are not anywhere near new. I remember looking at one a while back that used plain old SDRAM. That's a bit hard to find today; it's often actually cheaper to get DDR2 today than DDR. In higher capacities as well.

Of course; this brings up the question of 'what's the use?' - Seriously, unless you have something weird going on, just install a 64 bit OS and put the extra memory directly into your system.

Looking at all those benchmarks; except for the most artificial disk-thrashers; benchmarks for all were within 1-10%. Lost looked to be less than 5% difference between max and min.

Re:Performance issues...

[Gilmoure]

Mac OS7 had a way of creating a RAM disk from installed RAM. Was sweet way to run Photoshop at top speed back then, along with the dedicated Photoshop video card. Such specialization back then.

Re:Performance issues...

[drsmithy]

For DDR2, maybe, for fixed disk usage? Still pretty good compared to spinning disks or even flash.

A relatively modest (by today's standards) 8-spindle RAID10 (eg: a Dell PE2950's internal disk) will exceed 400MB/sec in sequential read operations.

Re:Performance issues...

[pipatron]

Of course, the Amiga could do this already in the eighties, and it could also keep its state during a reboot, you could even boot from it.

Re:Performance issues...

[TinyManCan]

But would the same 8 spindles handle the 20,000 IO/sec rate that this single drive manages? Not everyone is bound by the raw sequential transfer rate. In fact, almost no one is.

Re:Performance issues...

[fyonn]

ahh the amiga, a mighty computer! I still have an A1200 (with 50mhz '030 and '882) in a cupboard. no idea what to do with it, seems a shame to throw it away, but it's not worth anything!

and yes, the amiga was a computer massively before it's time. why didn't it win?

Re:Performance issues...

[Clover_Kicker]

This disk would be fantastic in a dedicated compiler box.

Compiling is usually I/O bound, and the intermediate files and compiled programs are ephemeral. Who cares if the power goes down and you lose everything, just recompile and you're back in business.

Gentoo here I come!

Re:Performance issues...

[Wonko the Sane]

This disk would be fantastic in a dedicated compiler box.

Compiling is usually I/O bound, and the intermediate files and compiled programs are ephemeral. Who cares if the power goes down and you lose everything, just recompile and you're back in business.

Gentoo here I come!

I'll save you a few hundred dollars:
mount -t tmpfs -o size=2G none /var/tmp/portage

Make sure you have enough swap space just in case you fill up your ram, but normally this will all but eliminate the disk IO delays.

Re:Performance issues...

[drsmithy]

But would the same 8 spindles handle the 20,000 IO/sec rate that this single drive manages?

Of course not, but I was replying to someone saying that a 200MB/sec controller bottleneck is "pretty good compared to spinning disks".

Re:Performance issues...

[Gilmoure]

Never used DOS. Took Intro to DOS as a Freshman and, since no intro to Mac, Desktop Publishing I on Mac. 4 weeks in, we were still renaming and moving files in DOS and were working on a newspaper in DTP class. Dropped DOS class.

Re:Performance issues...

[Gilmoure]

I still have my Quadra 650 (33Mhz 040), maxed out on RAM/HD/Vidio card. Machine's never died on me and I just can't bring myself to toss it. Hell, it fires up PShop 2.5 faster than my Dual G5 tower.

Good old hardware shouldn't be tossed. If I can find enough desk space at home, I'll set it up for my daughter to play Lemmings and Marathon I.

Re:Performance issues...

[Firethorn]

As another noted, Amiga could do this as well. For that matter, DOS could as well.

16 bit machines couldn't address much, so had to do some tricks.

Re:Performance issues...

[Firethorn]

Between any proper UPS and the battery these things have, even then you're unlikely to lose information.

Still, doesn't compiling normally involve more reads than writes? Because in that case, a Flash SSD would be cheaper and just as effective.

Re:Performance issues...

[Firethorn]

Pile these up, given that this maxes out at 64GB(?) per port, I don't think that, for the capacity, spinning disks will beat this thing. At least without going to enterprise level stuff, but at that point there are enterprise level RAM disk emulators.

Re:Performance issues...

[ya really]

Though it's not free, this will give you a ram drive on windows xp for all those with more than 3.25gb of ram. I use it for my page file and for the web browser cache.

Re:Performance issues...

[Firethorn]

Interesting, but even though I have TWO machines with 32bit OSes and 4 gigs of ram(and I have the modules to expand one to 6GB), meaning I'd be able to create a 1GIG ramdrive(due to 1Gig video card), and a 2.5 Gig in the other, I don't think that it'd be worth the $35.

Re:Performance issues...

[linhares]

I want to get a laptop with 64GB of ram. Price is not an issue. Can you tell me where to find that?

Re:Performance issues...

[Firethorn]

Where are you going to get a laptop with the external SATA ports to use this device?

Hmm... Here you go:

http://cgi.ebay.com/NEW-Lenovo-Thinkpad-X200-Laptop-64GB-SSD-BT-12-1-P8400_W0QQitemZ300287618088QQcmdZViewItem ;)

Re:Performance issues...

[linhares]

I don't mean SSD or other disks. I ram sheer giant 64GBs of RAM. (I'm developing a memory-hungry app, by the way).

Re:Performance issues...

[PipsqueakOnAP133]

That actually wasn't all that special.

What was REALLY special was that back then, the Powerbooks can make the ram disk, then select them in Startup Disk and boot off it!

Powerbook 2400c, booted to a desktop off a ramdisk copy of Disk Tools in 2 seconds flat.

There are better alternatives.

[BabyBrumak]

At the university we've been playing around with a new storage system that has ridiculous capacity and speeds. It uses 120 1TB drives to create a striped 100 TB storage system with redundancy. Fast writes are made through the interface directly into the RAM for one of the control computers. These computers each have 32GB of RAM and write the data onto the hard disks over a longer time than the initial dump to memory. You can get effective transfers up to 10 Gbps through the fiber channels. The best part is that this system uses off-the-shelf components with a custom build of Linux to handle IO. It appears as a standard high capacity network drive.

Re:There are better alternatives.

[rzei]

Offtopic but how do you practically backup 100TB and keep many backups? Hopefully you aren't using Seagate drives ;) Does your project have a site?

Not new

[gweihir]

You can get devices like this for at least a decade, possibly much longer. The primary advantages of FLASH are low price, high reliability and battery-less non-volatility.

Re:Not new

[GameMaster]

Most of what you said is right, but I don't know if I agree with "high reliability". Flash suffers from a limited number of writes per memory cell that RAM doesn't suffer from. Assuming the RAM disk is made with decent productiojn quality, it should be much more reliable, in the long term, than a Flash drive. The difference between this device and the previous ones is that this one is priced for the consumer or low end server market where most previous ones were priced for the enterprise market.

Re:Not new

[gweihir]

Given that current wear-leveling, FLASH durability and far better ECC, I think that FLASH may have reached DRAM reliability. The important thing is not for it to live forever, the important thing is to detect when it cannot reliably write anymore. If you factor in that DRAM has bit-errors as well and that typical DRAM ECC can only correct one bit-errors and detect 2 bit-errors (per 32 bit though), and in addition that power-failures over some minutes will wipe the whole thing, FLASH reliability could already be much, much higher, especially when compared on cost.

On the more practical side, several people are currently doing long-term FLASH overwrite experiments with quality FLASH. None have reported failures so far. The days of 10'000 overwrites, no wear-kleveling and no powerful ECC are over, at least for quality products.

Two Bays

[zogger]

It takes one bay for the RAM drive with built in battery, but they could have an additional full drive bay expansion battery with a jumper cable connector to give it a much longer hold time with the machine shut down.

Re:No ECC. & slower then a motherboard with 64

[haystor]

I'd use this for some of the math research I do. I currently run something for hours at a time and could save to disk every couple hours. If the computer were interrupted mid cycle, it would be garbage anyway.

Random reads *and* writes would be super fast. I imagine some guys working with video would love this.

I also wouldn't mind having my games on RAM so everything loaded 20x faster.

Re:solid state

[Stewie241]

What does battery filure leading to total data loss have to do with it being a solid state device?

Oh, the joy of RAM drives

[bmwloco]

When I was a poor, poor college student (not much different now, actually) I had a RAM drive in the 386 I cobbled up. Even a 10meg hard drive was prohibitively expensive, but I had memory. So, I set up a 3.5" floppy to create a ram drive on boot up, copy Procomm plus and text.exe into the RAM drive. Then I could log into GEnie and copy data. Typing "exit" saved my data and allowed me to power down. Oh, the job of batch files...

Re:ATTN: SWITCHEURS!

[Hordeking]

If you don't know what Cmd-Shift-1 and Cmd-Shift-2 are for, GTFO. If you think Firefox is a decent Mac application, GTFO. If you're still looking for the "maximize" button, GTFO. If the name "Clarus" means nothing to you, GTFO.

Bandwagon jumpers are not welcome among real Mac users. Keep your filthy PC fingers to yourself.

Did you just find the "Any Key"?
I have a key combination for you. It'll tell your computer to work for you while you just sit there and collect money. Just press the Alt key, Ctrl key, and Delete key all at the same time!

mOOF

too many pins for $200

[emj]

Not knowing anything about memory buses really makes this hard to comment, but I'm pretty sure you need lots of pins to interface memory.. And you need space on the motherboard, also such a motherboard wouldn't be massmarket. All these things costs lots of money (and is available for large quantities of money).

This solution is beautiful because it's simple and inexpensive ($200 for 16GB), max costs is 8x4GB modules.

Though I think you have a point that there should be a new class of I/O interface that can handle these types of disks better. Perhaps as tiered memory, where some of it is slower than the rest.

price is $370

[emj]

$200 is just for the memory, sigh..

ram disk cold boot

[emj]

Cool thing is you can cold boot memory from these as well, so back when you needed to reboot your computer because of memory corruption you could boot really fast by using ram disk..

Re:ram disk cold boot

[X0563511]

boot really fast by using ram disk. ... which was conveniently erased when you rebooted. Sounds useful.

Re:ram disk cold boot

[emj]

No, you actually booted from ramdisk as long as you didn't shut it off for more than ~5 seconds.

Re:ram disk cold boot

[X0563511]

Don't you risk silent corruption/loss when you do that?

Re:solid state

[SBrach]

I don't think "solid state" means what you think it means.

Re:solid state

[networkBoy]

Seems simple for me really,
I boot from SSD and all my files are on SSD. This is scratch disk for the gob of small reads and writes my system does (connected to some lab hardware), when a job is done it .tars and writes back to the SSD. I was going to re-write(engineer) the app to use memory caching but this would be simpler, and so is what I'm going to do.
-nB

Re:No ECC. & slower then a motherboard with 64

[AVee]

Which ofcourse isn't really an option anymore when your talking about a server which is not constantly being monitored by someone close enough to hit the button or for any site with more the a few servers.

It's fine however as boot disk containing static data. You know, the kind of stuff you'd restore from a disk image anyway. If you want fast booting servers, or something to serve your static data really fast this might be an option.

Re:ATTN: SWITCHEURS!

[hierophanta]

well i pass your test, but i'll go ahead and still GTFO. MACs are for trolls (evidently)

Back in the day

[celtic_hackr]

When PCs were still young there were add-in cards for memory, I still have some of these dinosaurs laying around. Certainly it makes sense that daughter boards could be made to create RAM disks that just plug into an empty slot on the PC, or in a case with some really wide and fast pipe for a laptop.

Solaris ZFS could use this

[tbuskey]

Most of the posts say what's the point, I'd rather stuff it into my regular RAM.

ZFS could use something like this for the ZIL/etc if it's faster then existing drives.

Battery backup means it keeps the content. Can it flush the ram to flash when the power goes out? Then load it back into its ram when the batt. is back up. That'd be perfect! No wear on the flash + protection from power outages.

PCjr

[eples]

Hey, my PCjr had a RAM disk. Worked pretty damn well too. I guess good ideas never stop working!

As Long As It Has Power

[aoheno]

DRAM loses it's data when power is switched off. Battery backups and other fail-safe features designed to keep data persistent in DRAM based disks might be considered too risky.

If your computer is off and the battery backup fails - you will have a freshly 'wiped' hard disk within a few seconds/minutes.

Which is why DRAM 'disks' are only really good for temporary data - like your download of the specification for a DRAM 'disk'.

Still just pretending....

[AcquaCow]

This is still just RAM pretending to be a hard drive, SATA in this case...

Let me know when they produce a NAND based flash memory card that plugs directly into the PCI bus for ridiculous IOPS...

Oh, wait... they've already done that... it's called FusionIO...

FusionIO is the only real SSD solution right now if you need to push > 80k IOPS.

These aren't the cheapest things out there, but if you run large databases and need to get your database to actually be per formant... an array of these is the way to go.

A bunch of ram pig-tailed off of SATA is just horrifically slow in comparison...

-- Dave

Re:Still just pretending....

[afidel]

What kind of IOPS are those for Fusion IO? I can get about 40K from an Intel X25-e @4KB 100% random writes which impressed the hell out of me and they are closer to $20/GB than the $30/GB I've heard quoted for the Fusion IO cards.

Re:Still just pretending....

[AcquaCow]

http://www.fusionio.com/Products.aspx

They have a test sheet there...

Their #s quoted are for 1KB and 512KB, but they show 140K IOPS random read @ 1KB...

We tested an array of 5 of the 320GB cards and tested with MS's SQLIO app and pulled some fantastic #s...

Re:Still just pretending....

[afidel]

5 cards? Are you doing software RAID-5 or something? I would think parity calculations would be crippling at the kind of rates you could get out of 5 of those cards.

Not the cheapest things out there....

[conureman]

You mean the FusionIO {drool} isn't the most expensive thing out there?

use

[matushorvath]

Ideal for developers who work on huge projects. You can put all your .o files and the compiled binary on the RAM disk and the compilation/linking performance should be marvelous. And if there is a power failure, you can always just recompile everything. I will seriously consider getting one of these...

Most tasks are not I/O bound

[symbolset]

At least they aren't until you put 16 cores in one box and share the I/O across all the cores, running 40 p2v servers.

Then all tasks are I/O bound. 1/16th of a 4Gbit SAN connection is 32MB/s. 1/40th is 10MB/s in practice. That's sad. That's 1980's I/O.

Why bother?

1. Money.

2. Desire to excel.

Why bother to do anything, really?