Beta
×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

OCZ Wants To Cache Your HDD With an SSD

Soulskill posted more than 2 years ago | from the wonder-if-this-will-cache-on dept.

Data Storage 189

sl4shd0rk writes "OCZ is coming out with Synapse Cache; an SSD cache for your hard drive. The SSD runs software that copies data into the cache from your hard drive as you work with it. The data sits on the SSD until it gets less activity and gets flushed to the hard disk. Aside from boosting your IOPS to 10k/75k (read/write), the SSD also supports AES encryption, SMART and TRIM."

cancel ×

189 comments

Sorry! There are no comments related to the filter you selected.

So... (1)

Jailbrekr (73837) | more than 2 years ago | (#37493490)

Its ZFS for Windows then?

Re:So... (5, Funny)

Bill, Shooter of Bul (629286) | more than 2 years ago | (#37493542)

Exactly! ....In the same way that a meatball is a golf ball for those playing the game of spaghetti.

Re:So... (2)

TheSpoom (715771) | more than 2 years ago | (#37493662)

That's simultaneously the best and least sensical analogy I've read on Slashdot in a long time. Kudos.

Re:So... (2)

Reber Is Reber (1434683) | more than 2 years ago | (#37494196)

Bravo sir.

Re:So... (2)

TorenAetonra (531269) | more than 2 years ago | (#37494236)

This! Yes!

Re:So... (1)

operagost (62405) | more than 2 years ago | (#37494302)

BadAnalogyGuy has a contender to deal with.

Re:So... (1)

ToasterMonkey (467067) | more than 2 years ago | (#37493758)

Windows ReadyBoost would work about the same way as ZFS's l2ARC if it allowed using whole SSDs instead of flash drives. It's basically the same idea, a second level page cache that can be removed or fail at any time.

Why desktop PCs continue to be built with the 'one large spinning disk one small SSD for important stuff' design while desktop operating systems totally ignore the potential of using them as cache boggles my mind.. I swear it is a conspiracy to make people buy larger SSDs than necessary instead of a more sensible blend of cost/performance like what Sun was after.

Re:So... (1)

arth1 (260657) | more than 2 years ago | (#37493906)

Windows ReadyBoost would work about the same way as ZFS's l2ARC if it allowed using whole SSDs instead of flash drives. It's basically the same idea, a second level page cache that can be removed or fail at any time.

No, it's not the same. ReadyBoost provides a compressed paging file cache, it doesn't work as a cache between the OS and the HD.
In other words, it only gets what is written to the paging file, or pages that are dropped from memory. Combined with SuperFetch, ReadyBoost gives a good boost in starting applications, but it neither works as RAM (the most common misconception), nor does it increase disk write speeds.

That said, Windows already has an API for hybrid disks, with both SSD and HDD that does what this solution promises. So I fail to see what's new, except perhaps a more user friendly setup?

Re:So... (2)

Strange Ranger (454494) | more than 2 years ago | (#37494826)

Prepare to have your mind unboggled:

For most people, the OS has been fast enough for the last decade. Boot times happen once per day at the most, programs launch once or twice per day and reside in large amounts of fast RAM. Even games reside just fine in 12GB of DDR3 and run like a champ on mid grade video cards. Where the speed breaks down for home users is Photos and Video. Everybody and their mom has digital photo and video equipment that fits in a purse or pocket.
It's working with these files where the SSD shines. Forget OS caching. We want media files _initially_ read from disk to be FAST. I want to transfer 24GB of hi res images to my pc and build a Lightroom catalog all in 2 minutes so I can work on them.. (I'm still waiting.). I want hd video to load and take edits without a pausing every 3 seconds while the disk spins. Storage AND Caching for these types of apps is great on an SSD. I certainly don't want windows contaminating my sacred SSD with it's super high, low ROI IOPS.

Re:So... (3, Interesting)

hairyfeet (841228) | more than 2 years ago | (#37494830)

Well according to this answer [superuser.com] by a developer you CAN use an SSD for readyboost, its just isn't as straight forward and you can't use the whole drive. personally I've been avoiding SSDs until they get the bugs out as the experience from my gamer customers (who spent waaaay more than i would have for top o' the line SSDs) is that Jeff Atwood at coding horror is correct that SSDs should be judged on a hot/crazy scale [codinghorror.com] as while they are crazy fast the fail crazy often.

To me it isn't THAT they fail it is HOW they fail that has me avoiding them. With HDDs I can't remember the last time I had an HDD that failed without plenty of clear warnings something was up. Windows delayed write fails, or SMART errors, temp going nuts, there was ALWAYS a clear warning that there was trouble in HDD town. With both of the gamers there was NO WARNING with the SSDs, they just flipped the switch and....nothing. With the HDDs I was always able to get the data off before they bought the farm, minus a few bad sectors of course, but with the SSDs it was like they didn't exist, it was just...nothing.

so while using it as a cache (as long as the cache is ALWAYS backed up like Readyboost) sounds fine i really can't see recommending an SSD until they get the bugs out. you would have to spend all your time running back ups or RAIDing the drive constantly to remove the risk, and that is just more trouble than its worth. Besides with Superfetch and Readyboost if you have a large amount of RAM (and what geek don't right? hell even my netbook is gonna have 6Gb on it) then everything you use often is already preloaded into RAM so unless you boot daily i doubt you'd see much difference, as nothing yet beats RAM speed.

Re:So... (1)

GameboyRMH (1153867) | more than 2 years ago | (#37494474)

On this topic, does anyone know how to set up something like an unRAID cache drive [lime-technology.com] using only FOSS tools? bcache is the closest thing I've seen and it's not really the same thing either.

Re:So... (0)

Anonymous Coward | more than 2 years ago | (#37494534)

More like the L2ARC of ZFS, but a far sight from ZFS.

'coming out' ? Seagate already has it (1)

unity100 (970058) | more than 2 years ago | (#37493508)

And its called momentus. it has a ssd cache which keeps most accessed files in itself. how clueless was the poster ?

Re:'coming out' ? Seagate already has it (1)

characterZer0 (138196) | more than 2 years ago | (#37493546)

Is it aware of the filesystem and keeps most accessed files, or just the most accessed blocks?

Re:'coming out' ? Seagate already has it (1)

alphatel (1450715) | more than 2 years ago | (#37493616)

It holds the high usage data until it can be written to disk (cold data). It won't be useful if you're running SQL databases with 80gb instances, but would certainly speed up your everyday PC-type activity.

Re:'coming out' ? Seagate already has it (1)

rthille (8526) | more than 2 years ago | (#37494564)

Well, the database example really depends on how much of the 80GB database is "hot". Even then the flash-based cache will probably help thruput as smaller writes to the platters can be reordered and combined.

Re:'coming out' ? Seagate already has it (2)

Anthony Mouse (1927662) | more than 2 years ago | (#37494406)

Is it aware of the filesystem and keeps most accessed files, or just the most accessed blocks?

How is that an advantage? If I have e.g. a 5GB mail archive file which is my most accessed file, but I really only frequently access a few dozen MB of it which represent the most recently received mail, the performance will be much better if it will cache the frequently accessed blocks only and then use the rest of the SSD to cache other recently accessed blocks than if it uses up half the SSD to cache the whole mail archive even though I haven't accessed 7/8ths of it in the last two years.

Re:'coming out' ? Seagate already has it (1)

Anonymous Coward | more than 2 years ago | (#37493818)

Actually, the momentus only has a read cache. This one can cache both read and write.

Re:'coming out' ? Seagate already has it (1)

Hatta (162192) | more than 2 years ago | (#37494530)

What's the difference between this and just keeping your swap file on SSD?

Re:'coming out' ? Seagate already has it (1)

unity100 (970058) | more than 2 years ago | (#37494676)

the difference in between these is, you end up keeping often accessed small (4kb or similar size) operating system or program files in ssd. ssds perform over 40 times faster or more for accessing such files. the difference that results in speed is phenomenal. also it is noticeable for bigger program files.

Re:'coming out' ? Seagate already has it (1)

Hatta (162192) | more than 2 years ago | (#37494792)

Isn't that what would happen if you used your SSD for swap? Or is disk cache simply freed instead of paged out when RAM is needed?

Good idea, how will the implementation be ? (3, Insightful)

zaibazu (976612) | more than 2 years ago | (#37493514)

They need to make the controller logic bullet proof, Seagate had quiet some problems with their hybrid disks

Re:Good idea, how will the implementation be ? (0)

Anonymous Coward | more than 2 years ago | (#37493810)

Exactly. I replaced a Seagate Momentus drive I got in an Alienware laptop with one of the Seagate Momentus XT Drives. The XT drives are the ones with the SSD cache.

It provided zero improvement in anything. It went straight back to the store after a weekend and some benchmarks. Sounds good in theory but not in practice as I've seen it yet. Hopefully the OCZ drives actual benefit the user.

Re:Good idea, how will the implementation be ? (1)

arth1 (260657) | more than 2 years ago | (#37493994)

That's because it pretty much only caches "startup files" on the SSD, and the SSD part is ridiculously small. You get a boost for boot speed, not much speed boost otherwise (unless you happen to re-read the files that were used during startup), and likely a worse MTBF, because there are more parts that can fail.

DNW.

Re:Good idea, how will the implementation be ? (1)

deroby (568773) | more than 2 years ago | (#37494698)

I beg to differ, although I will agree that it depends a lot on your usage profile.

Personally I find that that things like booting and launching apps is REMARKABLY faster than most people sitting around me having 'ordinary' drives.
(eg Outlook & VS2010 open up in a fraction of the time it takes the others). That said, yes, 4Gb is kind of 'low'. When I ('m forced to) use Word for a couple of days, *that* will start up much faster after a couple of days, but then after a week Excel will be slower to start and vice versa...

Frankly, I find the software mentioned in the article much more looking like what eBoostr did, but then for both read & write. I liked eboostr before I got my XT, but then again I always had this nagging feeling : how can you be sure the cache is always uptodate ? I'm having this very same feeling with this OCZ thing.

They should have hooked it up hardware-wize IMHO : mobo SSD HDD, doing it via software is just wrong.

Re:Good idea, how will the implementation be ? (4, Informative)

fuzzyfuzzyfungus (1223518) | more than 2 years ago | (#37493822)

A best I can tell, this is simply a basic SSD that is shipped with a bundled OEM copy of "dataplex" software from these guys [nvelo.com] (nice clip art...) (Here is a presentation by them about their product [flashmemorysummit.com] .

The SSD itself is a Sandforce 2281-based MLC drive with 50% overprovision for redundancy. Unless they've really screwed the firmware, it should be just fine, though no word on how it competes in price with other drives of similar size.

The caching function(unlike the Seagate hybrid units) is simply software: Supports Windows 7, no BIOS goo or specialized SATA features required; plugs into the OS somewhere in the storage handling area and shuffles data between the main mechanical HDD and the designated cache SSD.

On the plus side, that should(at least conceivably) give it considerably higher-level knowledge of what the OS is doing with which to make caching decisions(unlike caching firmware, which only has the SATA commands to go on). On the minus side, it means Win7 only, and your storage system is Not the place you want potentially flaky code, so if they aren't on the ball, we could see some serious bluescreening and/or OS hosing going on....

Re:Good idea, how will the implementation be ? (2)

gknoy (899301) | more than 2 years ago | (#37494488)

Would that kind of software be available so that I could do that with my own SSD+HD? I currently have an SSD that I keep gaming stuff on, but I wouldn't mind repurposing it as a cache if I knew how. Is there an easy way to do this?

Re:Good idea, how will the implementation be ? (1)

fuzzyfuzzyfungus (1223518) | more than 2 years ago | (#37495054)

I'm not sure. ZFS as support for using an SSD or SSDs as cache in a larger storage pool; but the phrase "that I keep gaming stuff on" usually does not imply "I run Solaris/BSD/Linux with ZFS/FUSE". In Linux, btrfs either has, or is working toward, some sort of SSD optimizations, I'm not certain how close they are to ZFS'.

Most of the reasonably nice SANs and storage appliances have support for some similar caching thing, to RAM, SSD, or a combination; but "Buy a SAN and bootable HBA" isn't exactly a desktop cost saving move.

I don't know of any commercially-available standalone software packages equivalent to this bundleware stuff, and I don't think that even the server versions of Windows do anything like that in NTFS or their software RAID modes.

Intel has something somewhat similar; but it is tied to motherboards with the z68 chipset, so that is likely either already helping you, or of no use.

I'm assuming that, sometime shortly after this product ships, cracked copies of the accompanying software will start filtering out; but I don't know when that will be...

Re:Good idea, how will the implementation be ? (-1)

Anonymous Coward | more than 2 years ago | (#37493940)

They need to make the controller logic bullet proof, Seagate had quiet some problems with their hybrid disks

Yeah. At least they didn't have some loud problems. You stupid fuck.

Couldn't I just do this with a RAM cache? (1)

Greyfox (87712) | more than 2 years ago | (#37493534)

Modern operating systems do that automatically anyway, as long as there's free RAM. It'd probably be less expensive to add another 32-64GB of RAM to your PC, than it would be to buy dedicated hardware to do that job.

Re:Couldn't I just do this with a RAM cache? (1)

EdZ (755139) | more than 2 years ago | (#37493606)

A 64GB SSD is about £100 nowadays. 64GB of RAM (ignoring the price of a motherboard with enough slots in which to fit it) is closer to £320 (buying cheapest-of-the-cheap 2gb stocks).

Re:Couldn't I just do this with a RAM cache? (1)

hedwards (940851) | more than 2 years ago | (#37493674)

For servers that would possibly be worthwhile, but I'm not sure why a home user would need 64GB of cache. 2GB would probably more than enough. Folks needing more than that would probably just opt for a 64GB SSD and be done with it.

Re:Couldn't I just do this with a RAM cache? (2)

networkBoy (774728) | more than 2 years ago | (#37494174)

I use all of 16 gb of ram as an advanced home user doing video editing (by no means am I a pro).
That said, I use spinning disks for the video work because I amd rewriting blocks all the time with scratch and render files and while an SSD is faster, it's not enough to make up for the killing of the drive in short order (which I've already done once).
-nB

Re:Couldn't I just do this with a RAM cache? (1)

fnj (64210) | more than 2 years ago | (#37494340)

Who uses 2GB sticks to get 64GB anyway? You can get a 4GB high quality stick for $21.99 at Newegg [newegg.com] . That's $351.84 for 16 of them. That's £227 at today's exchange rate. You're right about the ratio, though. It's about 3:1 for price of RAM to SSD.

And you know what? I'd take the RAM in a heartbeat. It would have at least 10:1 better read rate, and even more advantage in write rate. It never wears out. And you can use it for anything; not just a hard drive cache. In fact, I'd say the sweet spot for a hard drive cache is maybe 8GB, not 64GB, if you optimize what it caches. Now you're down to $44, and there are reasonably priced desktop motherboards that will hold this much more than you want to have for other purposes. On my desktop system, /bin, /lib, /usr/bin, /usr/lib, and /etc put together in their entirety total no more than 1.3GB. If you put all that and other well selected stuff in RAM drives and mount them, just about all programs will load essentially instantly. Just stick with the hard drive with existing linux RAM caching for the data (all other partitions).

Re:Couldn't I just do this with a RAM cache? (1)

Mitchell314 (1576581) | more than 2 years ago | (#37493610)

Why would you want to cache your RAM to an SSD? :P

Re:Couldn't I just do this with a RAM cache? (2)

myurr (468709) | more than 2 years ago | (#37493644)

No. The problem with caching in RAM is that it is volatile so if the power fails you'll lose the changes. Where this SSD helps is in providing a speed boost over a magnetic hard drive but without the volatility of RAM. The RAM cache will still be used to provide a further speed boost, but when a program issues an fsync to make sure that all the data held in the RAM cache is flushed to a physical disc, it will be the SSD that is picking up the slack.

For a read heavy environment with lots of RAM that is hardly ever switched off then probably not. For more general applications then it should provide a healthy speed boost without the cost of going all SSD.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37493690)

The cost of going all SSD?
Have you priced SSDs recently?
You can get 100GB drives for ~$150. How much cheaper does it have to be?

Sure you might still want a few spinning drives for bulk media storage, but that can be in one machine in the house not all of them.

Re:Couldn't I just do this with a RAM cache? (1)

countertrolling (1585477) | more than 2 years ago | (#37493782)

How much cheaper does it have to be?

Lots... The prices will always be too high while people are willing to pay it, but the cost of production does not justify it. I have a very hard time believing all that monkey motion going on in a regular hard drive costs less to make.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37493890)

If people are willing to pay it, then by definition the price is not too high.

I remember paying way more for HDs than $1.5/GB.

Re:Couldn't I just do this with a RAM cache? (1)

Dunbal (464142) | more than 2 years ago | (#37493930)

Eh, I remember when 64K cost close to $1 million including the cost of a new wing for it. Now get off my lawn.

Re:Couldn't I just do this with a RAM cache? (1)

operagost (62405) | more than 2 years ago | (#37494392)

64K? Luxury! In my day, we hadn't invented memory yet so we had to hire vagrants to stand in the computer room, each holding a card with a "0" on one side and a "1" on the other. We only had room for about 32,000 of these, for a measly 4KB. Each was tattooed with a memory location. When we needed to read the "memory", we had to shout out the appropriate drifter... I mean memory location. If the dirty hobo didn't answer within the refresh interval, he was beaten severely. High latency was not tolerated. The trick was giving each street person enough whisky to keep him sufficiently content, but not enough to put him to sleep. Sanitation was a real issue, but that's a story for another time. Back to work!

Re:Couldn't I just do this with a RAM cache? (1)

Bob the Super Hamste (1152367) | more than 2 years ago | (#37494504)

I remember paying about $33/GB for a HD. It was my first hard drive that I had bought separate from a computer. Granted this was back in early 98

Re:Couldn't I just do this with a RAM cache? (1)

tnk1 (899206) | more than 2 years ago | (#37493892)

Cost of actual production is not the issue, bringing up assembly lines for new stuff like SSDs so that production capacity for making SSDs is the issue.

Yeah, there's lots of fiddly things in a HDD, but we are really good at making them right now, and we also already have expanded production for them already. Startup costs are minimized. SSDs are still building up demand and production capacity is slowly coming up. It takes a lot of work to bring up a new process for a new product.

Re:Couldn't I just do this with a RAM cache? (0)

gad_zuki! (70830) | more than 2 years ago | (#37493954)

>You can get 100GB drives for ~$150. How much cheaper does it have to be?

I wish SSD enthusiasts would accept this simple fact: SSDs are not that reliable yet. RAM is reliable, barring a power outage.

If you want reliability, you at the very least need to buy Intel. Now your "cheap" drive isn't so cheap. We can't cheaply RAID them either so because the storage industry is so set in its way, they still haven't released much in the way of TRIM enabled RAID cards.

So lets say I wanted to do this in any way that was remotely reliable. I'd have to pay for a TRIM enabled RAID1 card and two 160gb SSDs, preferably intel. Now your $150 solution is more like $1500 solution.

Depending on garbage like OCZ is really asking for a fall. Business as well as consumers have been burned by the current crop of "fast but highly unreliable SSDs."

Unfortunately, we're probably years away from a cheap AND reliable SSD storage solution that can run for 5 years with the fail rate of a spinning disk. In the meantime I'm running expensive Intel SSDs and not worrying. I can't imagine selling OCZ to Joe or Jan consumer, inflating the price of their computers $200 for a mild speed boost. Not to mention laptops don't usually have a second slot for another drive. Either SSD all the way or spinning disk. Half-assed caching solutions have historically been big failures.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37494070)

Have you heard of this new thing called backups?

I don't care if the drive only lasts 1 or 2 years. By then a bigger, faster, cheaper one will be out and I will buy that.

I have backups if I use SSDs or HDs or clay tablets, so this changes nothing.

Re:Couldn't I just do this with a RAM cache? (1)

waives (1257650) | more than 2 years ago | (#37494334)

Yeah, because restoring from backup is so much fun.. I've used OCZ drives in both a desktop and a laptop, in each case they failed within 6 months. While the speedup was awesome, it was not worth the constant pain of repairing serious filesystem corruption every time I had a power failure or hard reset.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37494370)

Sounds like you need to fix your backup strategy. It is not fun, but once a year or two is no big deal.

I have not seen filesystem corruption, but at that point would get another drive and toss that one out. No point in having that happen more than once. I am not using an OCZ branded drive though.

Re:Couldn't I just do this with a RAM cache? (1)

waives (1257650) | more than 2 years ago | (#37494520)

Oh, my backup strategy definitely needs work.. in the first case I had none at all.

These were two separate drives purchased a year apart BTW though.

The main problem was not the final failures but more that nearly every time I lost power I encountered corruption.. and if I put off fixing it that led to crashes requiring hard reset -> more problems.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37494924)

I would have been returning the disk the first time that happened and I was sure it was not an OS issue. Like writeback being turned on for an ext3 volume.

Re:Couldn't I just do this with a RAM cache? (1)

myurr (468709) | more than 2 years ago | (#37494126)

So that's ~£1,500 per TB. Yet you can get a 1TB drive for less than the 100GB SSD... That's a HUGE discrepancy for bulk storage where the speed of retrieving every last byte doesn't matter.

This technology has its place for now and provides a welcome speed boost for those unable or unwilling to invest in a full SSD solution for their bulk storage needs. Going all SSD is the best solution for some people; going all mechanical is best for others; and there's a whole range of people with needs in between each extreme.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37494286)

Put the mechanical drives in another machine and use that for dedicated storage. All the other computers can have nice fast SSDs.

Z68 + msata (0)

Anonymous Coward | more than 2 years ago | (#37494496)

My power never fails (no I'm not THOR, just a bit rough) not in 10 years. Absolutely not a concern. But excuse me, isn't this already implemented on the motherboard level with z68 and Msata ... or other software & any SSD.

Re:Couldn't I just do this with a RAM cache? (1)

NitroWolf (72977) | more than 2 years ago | (#37493654)

Modern operating systems do that automatically anyway, as long as there's free RAM. It'd probably be less expensive to add another 32-64GB of RAM to your PC, than it would be to buy dedicated hardware to do that job.

This doesn't help you when you need:

1: Access to a file or files totaling more than 64 GB (or however much ram you have for caching)
2: Files not yet cached

Mostly I'm thinking games is where FS caching is less than optimum, since some of the resource files can easily exceed the amount of system memory available and if you like being competitive, the faster you load the game the better... and the FS cache won't help you there.

You also run into the problem of most motherboards don't support more than 8 GB (for older MB's), 16 GB for semi-newer MB or 32GB for some of the newest boards. Only a select few will support more than 64GB and those are the expensive boards, but to be fair most anyone considering an SSD purchase and/or 64 GB of RAM is likely to have a board like that anyway.

I run 24 GB on my system, and I don't see the FS cache filling up available RAM like I think it should. I'm not sure if that means I'm using less than 24 GB of data regularly (unlikely, since I play a lot of games and do large photo work), or the FS caching isn't designed around caching that much data and kind of tapers off after 8 or 10 GB.

Re:Couldn't I just do this with a RAM cache? (0)

Anonymous Coward | more than 2 years ago | (#37494206)

It also doesn't help when your bottleneck is transactions/second, not blocks/second. For example, your application wants to be sure the written data and metadata are atomically persisted before continuing to the next step. Having a persistent write-through cache would be great for this. The application (and OS) think the write is done, and it can happen at leisure for the big HDD, because it's already written to SSD. This is similar to how high-end RAID controllers have battery backup and/or flash backup for their RAM buffers, so they can acknowledge writes before they are committed to disk.

I wonder if we'll ever see an SSD-based cache sold as a SATA intermediary: host controller cabled to cache, cache cabled to drive, and the cache presents itself as the entire drive with non-volatile writeback cache capability.

Re:Couldn't I just do this with a RAM cache? (1)

MikeURL (890801) | more than 2 years ago | (#37494444)

Shouldn't the killer app here be the ability to enable instant on? Since an SSD can remember its state I'd think that it should be possible to take the power up time way down. Granted they'd have to get the motherboard, SSD, operating system and video card all working as a team but should this really be all that hard?

It seems to me that an HDD paired with an SSD should have enabled consumer level 'instant on' a long time ago. I must be missing something because it seems like a huge selling point that would be pretty easy to implement.

Re:Couldn't I just do this with a RAM cache? (1)

Baloroth (2370816) | more than 2 years ago | (#37493716)

Umm, are you aware just how expensive 32GB of RAM is? It isn't quite as pricey as I was expecting, Newegg has 24GB of DDR3 for as low as $160 (holy shit RAM is getting cheap), but that assumes you have 6 RAM slots, and I've not seen a consumer board with more than 4. If you want 8GB DIMMS (for 32GB on a consumer board), you're looking at more like ~$800 USD for 4. On the other hand, a 120GB SSD is around $160. Build a 64GB into a HDD, and it's still cheaper than anywhere near that in RAM.

The whole idea is to have a gradient for caching. RAM for stuff you need really fast, an SSD for stuff you often use and want pretty fast but don't need always, and an HDD for everything.

Re:Couldn't I just do this with a RAM cache? (1)

h4rr4r (612664) | more than 2 years ago | (#37493844)

6 ram slots is not that uncommon on triple channel memory boards. Here is one http://www.newegg.com/Product/Product.aspx?Item=N82E16813157163 [newegg.com] it will set you back an entire $155.

Re:Couldn't I just do this with a RAM cache? (1)

Baloroth (2370816) | more than 2 years ago | (#37494122)

Thanks for pointing that out, I hadn't seen one of those before (but I use almost exclusively AMD boards, so I guess I wouldn't.) But note that even that board supports a max of 24GB of RAM, so you couldn't necessarily even get 32GB, even with 6 slots.

Re:Couldn't I just do this with a RAM cache? (1)

ByOhTek (1181381) | more than 2 years ago | (#37493886)

6 GB consumer boards have been around for about as long as Intel's i7 series of processors (2 or 3 years)?

But, yeah, that's pretty inexpensive.

Re:Couldn't I just do this with a RAM cache? (1)

KingMotley (944240) | more than 2 years ago | (#37494294)

The performance chipset for intel is still the X58. All X58 motherboards have 6 slots. The fact that you've never seen a consumer board with more than 4 says taht you've never actually looked at Intel at all. Additionally, while not a typical consumer board, here's one with 12: http://www.evga.com/products/moreInfo.asp?pn=270-WS-W555-A2&family=Motherboard%20Family&series=All%20Motherboards&sw=5 [evga.com]

EVGA's dual 2011 board due in January 2012 should have 8 slots as well.

Re:Couldn't I just do this with a RAM cache? (2)

Ex Machina (10710) | more than 2 years ago | (#37493730)

Quick Google says:
64GB SSD = $86
64GB RAM = $2000

Re:Couldn't I just do this with a RAM cache? (1)

fnj (64210) | more than 2 years ago | (#37494382)

Maybe you could join the 21st century sometime. 64GB of RAM costs 16x$21.99=$351.84 [newegg.com] , not $2000.

Re:Couldn't I just do this with a RAM cache? (0)

Anonymous Coward | more than 2 years ago | (#37494714)

Quick Google says:
64GB SSD = $86
64GB RAM = $2000

Quick Newegg says:

64GB SSD = $95
64GB DDR3 = $352

Re:Couldn't I just do this with a RAM cache? (1)

timeOday (582209) | more than 2 years ago | (#37493746)

RAM cache is not persistent. That means when a user writes to a file, the system cannot (usually) return immediately after it's copied into cache; it must wait until the data is persistently stored. Thus slow persistent storage (HDD) is still a bottleneck even with lots of cache. (If you are doing non-blocking writes to a 64 GB non-persistent cache you could lose the whole system in a crash! Simply flushing that cache to disk could take an hour - does your UPS last that long?)

Also, the amount of disk reads necessary to fill a 32-64GB cache in the first place is huge. It would take hours for the system to fully warm up after being booted, because the first time you did anything it would still be HDD slow.

The problem I see with this is I would rather just choose what to put on HDD vs SSD manually. Most of my huge files are TV recordings that are written once sequentially, then read once sequentially, then deleted. That sort of activity tends to ruin cache by filling it with bulk data that will never be needed again. So having an SSD system drive and a HDD media drive is a pretty simple and effective separation.

Re:Couldn't I just do this with a RAM cache? (1)

networkBoy (774728) | more than 2 years ago | (#37494250)

does your UPS last that long?

mine lasts 40 min.
I still don't trust cached writes.

Re:Couldn't I just do this with a RAM cache? (1)

fnj (64210) | more than 2 years ago | (#37494526)

You have inaccurate knowledge of how caching works. Normally, write calls return as soon as the data is written to RAM, which is VERY quick. Writing back to hard drive then occurs in the background. Isn't multi-tasking great? You have to take special measures in your file I/O calls if what you want is to not have the system return until the data reaches the hard drive. That is very seldom appropriate or necessary, though you can do it.

It will never be necessary to wait an hour for writeback because the cache is writing back all the time in the background. Do you really think your apps are going to get 64GB ahead of writeback?

Your use case may make a mess of a STUPID cache, but there is no reason why a cache needs to be that stupid. In fact in the case described, those files should be stored on a partition with no caching at all, or only very minimal caching.

Z68 SRT does it better (0)

Anonymous Coward | more than 2 years ago | (#37493586)

Intel already has a tech that does this on Z68 boards and it works amazingly well.

Re:Z68 SRT does it better (0)

Anonymous Coward | more than 2 years ago | (#37493838)

It's a pretty big stretch to claim it works "amazingly well". If you have a (very expensive) Intel branded SSD, then yes it works pretty well, though I wouldn't go so far as to call the performance improvement amazing. However, if you have a different brand of SSD, you're potentially in for a world of misery. The management is partially handled by Intel software, and that software assumes your SSD supports the exact feature set of Intel SSDs. If it doesn't (most don't), then you'll find yourself wading through endless UEFI updates, driver updates, power management settings and registry changes trying as you desperately try to hem the never ending tide of BSODs. Look around on Z68 and SSD support forums, I am not the only one who felt this pain. The last wave of driver and UEFI updates were a massive improvement, but it's still not nearly as stable as it should be.

Smart Response Technology (1)

DarkXale (1771414) | more than 2 years ago | (#37493600)

This is different from Intel's SRT present on all Z68 boards how? The basics is the same. You have a HDD - and a SSD which acts as a cache. At a glance, I'm guessing the sole difference is that SRT is motherboard managed, while Synapse is done... elsewhere?

Re:Smart Response Technology (0)

Anonymous Coward | more than 2 years ago | (#37493738)

Intel makes an SLC SSD specifically for SRT too. Is this OCZ one MLC? You want the additional lifespan of SLC here because a cache requires writing many many times more frequently than a storage drive.

Re:Smart Response Technology (1)

dc29A (636871) | more than 2 years ago | (#37493788)

Exactly! Everyone already has a Z68 board! No one needs this Synapse crap!
Oh wait ...

Re:Smart Response Technology (1)

blackicye (760472) | more than 2 years ago | (#37494742)

Exactly! Everyone already has a Z68 board! No one needs this Synapse crap!
Oh wait ...

At $300 - $500, do yourself a favor, go buy a Z68 board and a normal SSD, and throw in a CPU if you need it too :P

Reliability (1)

Anonymous Coward | more than 2 years ago | (#37493604)

Ah, so OCZ finally worked out how to make a reliable drive... they just keep the entire content elsewhere.

Not a novelty (2)

c0l0 (826165) | more than 2 years ago | (#37493614)

Intel is doing the very same thing on their most recent "enthusiast" desktop chipsets.

For systems using the Linux kernel, there are software implementations of the very same block-level-caching-concept available - one I stumbled over is http://bcache.evilpiepirate.org/ [evilpiepirate.org]

Re:Not a novelty (1)

ToasterMonkey (467067) | more than 2 years ago | (#37493912)

Intel is doing the very same thing on their most recent "enthusiast" desktop chipsets.

For systems using the Linux kernel, there are software implementations of the very same block-level-caching-concept available - one I stumbled over is http://bcache.evilpiepirate.org/ [evilpiepirate.org]

... Make no mistake, this should be the job of the operating system. Solaris has zfs l2arc, and Windows has ReadyBoost that is similar. Windows and Mac systems both ship with the big disk little SSD pattern. MS and Apple need to get off their asses and ship 'SSD as cache' software like Sun did. Only reason I can think of for not doing this is patent disputes, some angle that makes this not feasible for desktop use, or an intentional long term strategy to drive SSD costs down by poising them as spinning disk replacements rather than complements. Ok, for Apple that last one actually makes sense.. but Microsoft? They already have ReadyBoost.. I don't get it.

rewrite ? (0)

Anonymous Coward | more than 2 years ago | (#37493618)

my wife asks ...What kind of wonder ssd can handle so much rewrites? what kind of wonder ssd is faster than ram?

Reports to be in the 300-500$ range (1)

Osgeld (1900440) | more than 2 years ago | (#37493708)

Good god why, cause they added the word cache to the label. That has to be wrong

SSD Cache and corruption (3, Interesting)

DigiShaman (671371) | more than 2 years ago | (#37493760)

Not sure I'm feeling the love for this concept. On the reads, sure. Nice idea. Writes however, not feeling the love. For whatever the reasons, PC hardware can lock up (CPU, video, motherboard, RAM etc) or because of buggy device drivers on the OS. In any event, how well can this device recover from a dirty-cache shutdown? What happens if the device just dies? Will I still be able to mount the HDD and recover data? It would be interesting to see how a journaling file system handles the abstraction of one volume read/written between two different drives. Were not talking about RAID5 here where you at least have parity data to recover from.

Re:SSD Cache and corruption (3, Informative)

fdawg (22521) | more than 2 years ago | (#37494082)

Obv I have no idea how OCZ plans on doing this, but I can tell you what a standard journaling fs does.

In any event, how well can this device recover from a dirty-cache shutdown?

Chances are this cache is transparent. The blocks translate to vblocks which map to physical blocks on the rotational media. A "dirty" block is a vblock which hasn't been committed to the physical block. However, this is transparent to the filesystem. When the system comes back up and the journal is replayed at say, some operation 10, and we find the relevant blocks for op 10 which happen to be vblocks in the SSD, the write is stable. It's a NOOP from the filesystem perspective.

What happens if the device just dies? Will I still be able to mount the HDD and recover data?

This is the same as a single, non-tiered, drive dying. Same semantics- cache is dead is equivalent to the drive being dead. That is to say unless the journal and superblock live somewhere else. IIRC ext2/3 keeps the initial copy of the superblock in a few places on the drive. Depending on which you can recover, you'll get a version of the filesystem (likely the one when you first created the fs, i.e. an empty fs). In short, pay attention to your SMART data, and always (ALWAYS!) backup.

It would be interesting to see how a journaling file system handles the abstraction of one volume read/written between two different drives. Were not talking about RAID5 here where you at least have parity data to recover from.

Most journals arent like NVRAM and don't follow the copy on write semantic. Journal replay is usually a data-loss event if all writes weren't stable before the replay. With that in mind, most volume managers (the original "VM"!) allow your fs to write to as many drives as the vm allows. This seems no different. But yeah, maybe they're doing something smarter here.

Re:SSD Cache and corruption (1)

Twillerror (536681) | more than 2 years ago | (#37494590)

Since the SSD is non volatile I'm not really sure what a system lockup would do. If the byte(s) writes to the SSD and the system fails it is still on the SSD. When the system boots up the thing would be considered dirty and would write it out to the hard drive. I'm sure they will have to reserve blocks of data for dirty bits... a byte can store 8 bits each bit representing a much large ...like 64k block in the cache. A 1 means dirty a 0 clean or vice versa.

The same basic thing has been around for SCSI RAIDs for a while. The PCI card has high speed memory with a battery. If you enable write ahead cache it writes into this memory and then returns to the OS. Then it write out to the hard drive. If the system fails upon boot up the card would still have the cache because of the battery and could recover.

For db servers I would probably turn off write ahead cache, but for a normal computer even if there is corruption it probably would be easily recoverable.

Fi85t!? (-1)

Anonymous Coward | more than 2 years ago | (#37493766)

since we made the And some of the

Do SSDs benefit from process shrinking too (1)

G3ckoG33k (647276) | more than 2 years ago | (#37493826)

While good it is not all there is to it. SSDs benefit from the same process shrinking as CPUs and GPUs do.

For fun I created a this table which predicts what processes will be available in the future based on numbers from Wikipedia. They are all in nanometer, and should not be trusted for anything beyond 2011 :)

1971: 9095.3066, 1972: 7865.412, 1973: 6801.8275, 1974: 5882.064, 1975: 5086.6737, 1976: 4398.8385, 1977: 3804.0144, 1978: 3289.6242, 1979: 2844.7913, 1980: 2460.1101, 1981: 2127.4467, 1982: 1839.767, 1983: 1590.9882, 1984: 1375.85, 1985: 1189.8034, 1986: 1028.9147, 1987: 889.7817, 1988: 769.4628, 1989: 665.4137, 1990: 575.4345, 1991: 497.6225, 1992: 430.3325, 1993: 372.1417, 1994: 321.8196, 1995: 278.3022, 1996: 240.6693, 1997: 208.1253, 1998: 179.982, 1999: 155.6443, 2000: 134.5976, 2001: 116.3969, 2002: 100.6574, 2003: 87.0462, 2004: 75.2755, 2005: 65.0965, 2006: 56.294, 2007: 48.6817, 2008: 42.0989, 2009: 36.4061, 2010: 31.4832, 2011: 27.2259, 2012: 23.5444, 2013: 20.3606, 2014: 17.6074, 2015: 15.2265, 2016: 13.1675, 2017: 11.387, 2018: 9.8472, 2019: 8.5156, 2020: 7.3641, 2021: 6.3683, 2022: 5.5072, 2023: 4.7625, 2024: 4.1185, 2025: 3.5616, 2026: 3.08, 2027: 2.6635, 2028: 2.3033, 2029: 1.9919, 2030: 1.7225, 2031: 1.4896, 2032: 1.2882, 2033: 1.114, 2034: 0.9633, 2035: 0.8331, 2036: 0.7204, 2037: 0.623, 2038: 0.5388, 2039: 0.4659, 2040: 0.4029, 2041: 0.3484, 2042: 0.3013, 2043: 0.2606, 2044: 0.2253, 2045: 0.1949, 2046: 0.1685, 2047: 0.1457, 2048: 0.126, 2049: 0.109, 2050: 0.0942, 2051: 0.0815, 2052: 0.0705, 2053: 0.0609, 2054: 0.0527, 2055: 0.0456, 2056: 0.0394, 2057: 0.0341

The point is that these numbers will help SSDs, too.

Re:Do SSDs benefit from process shrinking too (1)

Nom du Keyboard (633989) | more than 2 years ago | (#37494008)

While good it is not all there is to it. SSDs benefit from the same process shrinking as CPUs and GPUs do.

For fun I created a this table which predicts what processes will be available in the future based on numbers from Wikipedia. They are all in nanometer, and should not be trusted for anything beyond 2011 :)

1971: 9095.3066, 1972: 7865.412, 1973: 6801.8275, 1974: 5882.064, 1975: 5086.6737, 1976: 4398.8385, 1977: 3804.0144, 1978: 3289.6242, 1979: 2844.7913, 1980: 2460.1101, 1981: 2127.4467, 1982: 1839.767, 1983: 1590.9882, 1984: 1375.85, 1985: 1189.8034, 1986: 1028.9147, 1987: 889.7817, 1988: 769.4628, 1989: 665.4137, 1990: 575.4345, 1991: 497.6225, 1992: 430.3325, 1993: 372.1417, 1994: 321.8196, 1995: 278.3022, 1996: 240.6693, 1997: 208.1253, 1998: 179.982, 1999: 155.6443, 2000: 134.5976, 2001: 116.3969, 2002: 100.6574, 2003: 87.0462, 2004: 75.2755, 2005: 65.0965, 2006: 56.294, 2007: 48.6817, 2008: 42.0989, 2009: 36.4061, 2010: 31.4832, 2011: 27.2259, 2012: 23.5444, 2013: 20.3606, 2014: 17.6074, 2015: 15.2265, 2016: 13.1675, 2017: 11.387, 2018: 9.8472, 2019: 8.5156, 2020: 7.3641, 2021: 6.3683, 2022: 5.5072, 2023: 4.7625, 2024: 4.1185, 2025: 3.5616, 2026: 3.08, 2027: 2.6635, 2028: 2.3033, 2029: 1.9919, 2030: 1.7225, 2031: 1.4896, 2032: 1.2882, 2033: 1.114, 2034: 0.9633, 2035: 0.8331, 2036: 0.7204, 2037: 0.623, 2038: 0.5388, 2039: 0.4659, 2040: 0.4029, 2041: 0.3484, 2042: 0.3013, 2043: 0.2606, 2044: 0.2253, 2045: 0.1949, 2046: 0.1685, 2047: 0.1457, 2048: 0.126, 2049: 0.109, 2050: 0.0942, 2051: 0.0815, 2052: 0.0705, 2053: 0.0609, 2054: 0.0527, 2055: 0.0456, 2056: 0.0394, 2057: 0.0341

The point is that these numbers will help SSDs, too.

And you reach the atomic level when...?

Re:Do SSDs benefit from process shrinking too (2)

Rockoon (1252108) | more than 2 years ago | (#37494338)

The current trend in flash process size reduction is to trade off erase endurance for the additional area shrinkage in pretty much a linear fashion.

So while the smaller tech allows for larger capacity drives in the same form factor, for two drives of equal capacity, equal over-provisioning, but different process sizes, you are much better off (in terms of erase endurance) with the larger process size.

I believe the latest shrink cut the erase endurance by approximately half, while approximately doubling density. At some limit you end up with exactly 1 erase cycle (and if doubling/halving is a trend, this wont take all that many process reductions) and you have to provision the vast majority of the drive (reserving all but fractions of a percentage!) for "wear leveling" in order to pretend that you have something other than a WORM drive.

Is there or will there be Mac OS X support? (0)

Anonymous Coward | more than 2 years ago | (#37493878)

OCZ should know better than to throw their weight behind the most rapidly declining platform in the market.

Re:Is there or will there be Mac OS X support? (1)

amicusNYCL (1538833) | more than 2 years ago | (#37494148)

OCZ should know better than to throw their weight behind the most rapidly declining platform in the market.

Yeah, just look at the numbers plummet. [netmarketshare.com] Over the last 2 years, from September '09 until now, Windows went from 93.85% to 92.90%, so it dropped just under 1%. So in another 2 years, it will be, what, around 91% - 92%? Yeah, why would OCZ want to pin their hopes on a prospect as miserable as that? They should probably go for that 6% - 7% that OSX commands instead.

There are plenty of arguments to be had between Windows and OSX, but market share isn't one of them.

Wouldn't it be great if it could be a USER add-on? (3, Interesting)

wisebabo (638845) | more than 2 years ago | (#37493894)

I mean did you know many people have laptops that can take a 12.5mm tall HDD? But most people only buy a 9mm drive?

So it would be nice if OCZ (or another manufacturer) could make a very thin (3mm) card that would piggyback on top of the HDD. It would also have to a SATA drive connector to attach it to the motherboard and then a loop through cable to attach to the drive. In this way the end user could add a SSD cache to their existing laptop!

Is this feasible or am I missing something?

Re:Wouldn't it be great if it could be a USER add- (1)

hahn (101816) | more than 2 years ago | (#37494058)

I mean did you know many people have laptops that can take a 12.5mm tall HDD? But most people only buy a 9mm drive?

So it would be nice if OCZ (or another manufacturer) could make a very thin (3mm) card that would piggyback on top of the HDD. It would also have to a SATA drive connector to attach it to the motherboard and then a loop through cable to attach to the drive. In this way the end user could add a SSD cache to their existing laptop!

Is this feasible or am I missing something?

Heat? Battery life?

Re:Wouldn't it be great if it could be a USER add- (0)

Anonymous Coward | more than 2 years ago | (#37494974)

2.7W while in-use, 1.5W while idle for this drive. Not trivial, but it's not going to cause any thermal problems, and with the right kind of caching policy, it would probably be a net power savings due to spinning up the hard drive less often.

Re:Wouldn't it be great if it could be a USER add- (1)

Caratted (806506) | more than 2 years ago | (#37494350)

I just submitted it to my patent office, I'll let you know.

Unexpected Events (1)

Nom du Keyboard (633989) | more than 2 years ago | (#37493990)

And just how well does this extra level of complication recover from every kind of unexpected system shut down/BSOD/you name it in the consumer PC world?

Uh... (1)

Syberz (1170343) | more than 2 years ago | (#37494012)

It's called Smart Response Technology (SRT) and it was introduced by Intel as part of their Z68 chips.

Granted, OCZ's version has a few more bells and whistles, but it's basically the same thing.

Windows only (0)

Anonymous Coward | more than 2 years ago | (#37494072)

I presume?
i.e. useless.

Cache Your HDD With an SSD (1)

Siggy200 (721326) | more than 2 years ago | (#37494172)

A few months ago purchased a Seagate Momentus XT 500 GB Internal drive. The drive offers hybrid storage with Adaptive Memory technology, enabling the drive to deliver higher capacity and SSD-like performance. I am very satisfied with increased performance from the drive especially running Microsoft Flight Simulator FSX, very little pausing effect while 'flying' with the same settings used in a ATA serial drive I had before.

Re:Cache Your HDD With an SSD (2)

nabsltd (1313397) | more than 2 years ago | (#37494754)

I am very satisfied with increased performance from the drive

It's just your imagination. Check out these benchmarks [techreport.com] , or compare it to even more recent standard hard drives [techreport.com] .

It's hard to imagine how bad Seagate must have been with the design of a 3-1/2" drive that includes flash memory cache when it is regularly beaten in benchmarks by standard 2-1/2" drives (e.g., WD Scorpio Black) and trounced by other 3-1/2" drives. Since it costs a lot more (5x as expensive per GB as the WD Caviar Green), it's about the biggest loser ever made.

Why use flash eeprom for an SSD that caches? (0)

Anonymous Coward | more than 2 years ago | (#37494254)

What I don't get is why they're pushing SSDs that use flash memory for caching. Flash has a finite number of writes so eventually you hit the limit. There are other existing technologies like FeRAM, MRAM, and PRAM which although they have a lower density at present are more suited to the caching process due to unlimited rewrites. The lower densitys arent such a big deal when caching when for most 40gb would be enough, but the current prices of these technologies may be prohibitive.

DM-Cache (1)

Anonymous Coward | more than 2 years ago | (#37494270)

Build your own on Linux using DM-Cache

Oh, goody, more proprietary! (1)

macraig (621737) | more than 2 years ago | (#37494426)

Anyone care to place bets on whether that DataPlex software is locked to only work with OCZ SSD hardware? Anyone care to bet on how long it will take before an open source equivalent appears on SourceForge and negates OCZ's proprietary stunt?

Why SSD? (1)

Dcnjoe60 (682885) | more than 2 years ago | (#37494480)

What is the advantage of using an SSD in this configuration? Granted, if the power went out, the SSD would retain the data versus a standard ram cache, but then again, the hard drive isn't spinning to accept the data. Also, SSDs tend to wear out in a few years, even with error correction. Now, most likely, the system would realize that and default back to the HD without the SSD, but again, a standard cache wouldn't have that problem, or at least not as soon.

The only advantage I see is that an 128GB SSD is a lot smaller than 128GB ram, but I don't know if it has to be that way. With a ram cache being volatile, a cmos style battery can mitigate that problem, too.

So, I ask, again, what is the big advantage to using an SSD for this?

Re:Why SSD? (1)

Arlet (29997) | more than 2 years ago | (#37495012)

The only advantage I see is that an 128GB SSD is a lot smaller than 128GB ram

Also cheaper, lower power. and non-volatile. A battery wouldn't work, because DRAM needs to be refreshed, and with the size of the RAM, the current drain on the battery would be substantial.

Load More Comments
Slashdot Login

Need an Account?

Forgot your password?

Submission Text Formatting Tips

We support a small subset of HTML, namely these tags:

  • b
  • i
  • p
  • br
  • a
  • ol
  • ul
  • li
  • dl
  • dt
  • dd
  • em
  • strong
  • tt
  • blockquote
  • div
  • quote
  • ecode

"ecode" can be used for code snippets, for example:

<ecode>    while(1) { do_something(); } </ecode>