IBM Flash Memory Breaks 1 Million IOPS Barrier

alphadogg writes to tell us that IBM is claiming a victory on the flash storage front. Their new research project "Quicksilver" is claiming data transfer speeds of more than 1 million input/output operations per second (IOPS). "IBM said Quicksilver is two and a half times faster than its own SAN Volume Controller coupled with IBM's DS4700 storage. It would also be two and a half times faster than technology from Texas Memory Systems, which says it has the world's fastest storage with an IOPS rate of 400,000. "

IBM Flash Memory Breaks?

[Yvan256]

Then I'm not buying IBM flash memory, end of story.

Re:

[maxume]

Will Farrell called, he wants you to join his club.

Re:

[Windows_NT]

Does it run linux?

Yay!

[Anonymous Coward]

Does this mean I can wear out my flash drive more quickly? WOO!

Bit error rate?

[Toffins]

That's very fast. I wonder how low the bit error rate is.

Re:

[pixelpusher220]

More importantly, how durable is the memory?

Doesn't flash memory have a limit on the # of accesses before it starts to fail?

Re:Bit error rate?

[Surt]

It has a limit to the number of writes, the number of reads is pretty unlimited. The expected average lifetime is similar to a hard drive, and in some cases better.

Re:

[HTH NE1]

The bigger the flash drive, the more area in which to spread the wear, the longer it lasts.

Re:

[Hurricane78]

...the more you wear it down by using it. (eg. bigger files, more files used.)

Isn't that the whole point of a bigger drive?

Re:

[winphreak]

If that is true, I would hope this can help issues with SSD implementation. Sadly, I don't know as much as I should but it seems like it would help.
Numbers aside, wouldn't speed increases like this help to establish the possibility of SSD raid configurations?

My ignorant two cents.

Re:

[Heembo]

It has a limit to the number of writes, the number of reads is pretty unlimited. The expected average lifetime is similar to a hard drive, and in some cases better.

I don't believe that at all. I have read the opposite - that flash wear for high-write systems, like a server - is very rapid, in fact Dell is getting returns for flash workstations at a rate faster than any other computer. You might be right, can you post a few links?

Re:Bit error rate?

[Surt]

Obviously, there's not a whole lot of independent research out there, but here are some of the claims:

http://talkback.zdnet.com/5208-11408-0.html?forumID=1&threadID=39451&messageID=725468&start=0 [zdnet.com]

http://www.linuxdevices.com/news/NS7676844023.html [linuxdevices.com]

http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9007518 [computerworld.com]

Re:Big deal

[ppc_digger]

Which means it transfers 640 GB per second.

Re:

[sexconker]

I am, and it's 610.3515625.

Re:Big deal

[Anonymous Coward]

It only transfers 640KB per I/O operation, tops.

Well, 640kb should be enough for anybody.

Time to market?

[Naughty Bob]

They don't even commit to a date when this might be viable.

Given that current systems are 3 or more orders of magnitude slower than the stated amount, I'm pretty safe in saying that this announcement is meaningless outside of the lab. Kudos, but.... next!

Re:Time to market?

[spun]

If it's in the lab, "Next!" But if it's in the market, "Slashvertisement!" Good old Slashdot, where someone is always ready to shit in your cornflakes.

Re:Time to market?

[Yvan256]

Oh crap, I thought these were Raisin Bran.

Re:

[sir fer]

I don't wanna learn about it unless I can buy it!! ;o)

Re:

[winomonkey]

So, if Slashdotters are shitting in my cornflakes, who the hell is pissing in my cheerios?!

Re:

[COMON$]

We have been looking into the Texas memory solutions, they are very impressive but I too wonder when this might be a viable commercially sold solution.

Re:

[sexconker]

What company was it that put out (essentially) a DDR -> IDE board/card?

That thing was AWESOME.

Expensive per GB, but screaming fast.

Re:

[blackicye]

That would have been the Gigabyte I-Ram [anandtech.com]

I'm waiting for price/performance and benchmarks

[afidel]

While managing to achieve 1M IOPS is somewhat impressive, it's not hard to do for an optimal theoretical situation. Xiotech was showing 500,000+ IOPS from three of their new Emprise 5000 storage shelfs at Storage Networking World this spring, but it was all video and synthetic sequential reads. That same system would only pull about 20K IOPS on the SPC-1 real world benchmark.

Re:I'm waiting for price/performance and benchmark

[afidel]

I stand corrected, from the talkback link I followed a trail to an IBM blog with a LOT more details here [ibm.com], and this is the 70/30 SPC-1 benchmark numbers with cache disabled. This is freaking phenomenal performance! The storage is only 4TB, but if you put your logs, flashback, and temp tables on this beast and pinned your busiest tables in ram you would have a screaming OLTP database. I guess it's now just a matter of price, but a rack of x-series boxes with flash card's shouldn't be THAT expensive. Unless IBM asks for a crazy markup it should be affordable for most enterprises (ok, pretty much a given with IBM but still).

Re:

[clem.dickey]

Read the link carefully. It is not SPC-1, nor modified SPC-1. It is 4K IO in a 70/30 read/write mix.

It's not a fucking barrier

[Nimey]

It's the 1 MIOPS /mark/. If it was a barrier, you wouldn't be /able/ to break it.

Re:It's not a fucking barrier

[Aranykai]

Italics look like this. /This/ /just/ /makes/ /you/ /look/ /like/ /an/ /ass/.

Re:

[Yvan256]

Cut him some slack, his C64 can't do italics.

Re:

[Nimey]

Usenet-isms.

*Bold* /Italic/ _Underline_

But I wouldn't expect you to understand that.

Re:

[truthsearch]

/Usenet/ /is/ /dead/ [slashdot.org].

Slashdot confirms it, so it must be true.

Re:

[Neoprofin]

Just because I can type in 13375p34k doesn't mean I shouldn't be told it makes me look like an ass.

Re:

[omnipresentbob]

But I wouldn't expect you to understand that.

Of course not... Just look at his UID (not that mine is much better...)

Re:

[jebrew]

With a reply that *recognized* your command, /I'd/ have thought _you_ would have understood that *he* understood.

/yay, slashies!/

Re:It's not a fucking barrier

[Just Some Guy]

If it was a barrier, you wouldn't be /able/ to break it.

Yeah [wikipedia.org].

Re:

[santiagodraco]

It's a barrier if it's considered very difficult to break. It's used all the time in this kind of context. But you'd know that if you weren't just grasping for something to say that you could attach an insult to. A "mark" implies nothing to do with difficulty, it's just a level, or stage, or, well, a mark.

Re:

[sir fer]

What? Like teh Sound Barrier?? [wikipedia.org]??

soooooo...

[Taibhsear]

How does this translate into normal transfer speed units like MB/s? Otherwise I have no point of reference to tell if I am impressed or indifferent.

Re:

[Anonymous Coward]

More importantly, how many Libraries of Congress per lunar month is this?

Re:

[JJJK]

about 15946
...using google calculator, 10 TB for a LoC and 640 GB/s.

So... do you measure velocity in furlongs per forthnight?

Re:

[alx5000]

My car gets 40 rods to the hogshead, and that's the way I likes it!

Re:

[afidel]

It's not freaking 640GB/s, there's only one switch in existence that can do that much (Cisco 7000, the Brocade DCX-Backbone is the only other one that's close and it's 6.4Tb/s total per chassis). It's ~4.4GB/s, 1.1M peak IOPS * 4KB chunks...

Re:

[sir fer]

My car gets 40 rods to the hogshead yada yada yada

Re:soooooo...

[Hecatomb00]

How does this translate into normal transfer speed units like MB/s? Otherwise I have no point of reference to tell if I am impressed or indifferent.

All I know is it is fast. This is a huge win in my book. I am really tired of finishing before my standard hard drive can seek out my porn.

Re:

[elgatozorbas]

All I know is it is fast. This is a huge win in my book. I am really tired of finishing before my standard hard drive can seek out my porn.

Why don't you give it a head start?

Re:

[afidel]

You can normally assume that for that level of IOPS they are 4K blocks, so 4GB/s, pretty damn impressive as that's saturating 4*10Gb/s links.

Re:

[Anonymous Coward]

You kidding me? It can make the Kessel Run in less than twelve parsecs. That's smoke'n

Re:soooooo...

[TheRaven64]

IOPS are a normal measure for server hardware. It's the number of I/O operations per second the device can perform. Most server workloads, particularly database servers, require a lot of small I/O operations per second. With a single mechanical disk, it's pretty easy to work out the number. Pretend the time taken for each read is zero (it isn't, but it's really tiny so we can ignore it). Then take the reciprocal of the seek time. For a cheap disk, this is around 9ms. Google says the reciprocal of 9ms is 111Hz [google.com], so a 9ms seek time translates to 111 IOPS. Fast drives have average seek times down at around 4ms, which gives 250 IOPS. So, to put this in perspective, it's as many independent operations per second as 4,000 individual high-end disks, or almost 10,000 cheap ones.

By the way, for a workload with a lot of independent reads or writes you'd be surprised how slow a hard disk is. With a 512byte block (common on hard drives) you get a maximum throughput of around 50KB/s for a random access pattern on a cheap drive, going up to around 125KB/s on an expensive one. Even very cheap flash can do better than this, so for moderate sized databases (a few GBs) with a very heavy access load flash works out a lot cheaper.

Oh, and for reference each of the ops in this test was up to 640KB, giving a maximum of around 640GB/s data transfer.

Re:

[jebrew]

Those IOPS aren't spot on, you've got to take into account sequential vs. random. I've got a USB drive here that barely pushes 25 MB/s, and has a seek time of ~6ms (old ide in an old usb case...mostly for beating with a stick), yet I can get ~2000 IO/s using sequential reading.

Otherwise, that's a good explanation.

Re:

[Amouth]

he is damn close for random iops though.. almost everything does sequential well.. as there is no seek time past the start.. so flash vs spindle is pointless.. where spindels hurt is random..

he was doing random - and was close enough for the net

Re:

[jebrew]

Good point. There are a lot of factors to take into account. I just felt like his post may have been a little misleading without specifying purely random I/O.

Re:

[owlstead]

Anyone that has done some downloading from a news server and par2-ing and unzipping at the same time can relate to that 50KB/s. Or copying multiple folders at the same time.

I really really will buy a fast SSD once they become available, and I am thinking about buying one or two of of these new WD velociraptors as well. Current hard drives suck. They are slow, noisy and still a bit unreliable (even though they seem to be *much* better than days of old).

Re:

[basil montreal]

MB/s is only a measure that's meaningful for sequential reads where the data can be prefetched. Most enterprise storage is based on applications that read randomly all over the disk (like databases and email servers). The benchmark measurement for this type of application is in the number of operations you can do per second. A single hard drive spindle can do between 80 and 150 IO/s, which would generate the number of IO/s times the size of the IO blocks per second.

Re:soooooo...

[TopSpin]

How does this translate into normal transfer speed units like MB/s? Otherwise I have no point of reference to tell if I am impressed or indifferent.

I'll try to help.

MB/s is a measure of IO throughput. Often this isn't the most relevant figure for 'enterprise' storage. Certain applications do a lot of random access IO so IOPS becomes more important than throughput.

Today a typical desktop disk is capable of about 100-150 IOPS. That's a rule of thumb range that varies based on operation size, cache, etc. It works pretty well usually. You can aggregate disks and get almost linear scaling; 12 disks, for instance in a device like this [cdw.com], will give you a maximum of 1200 IOPs, roughly. A common USB Flash device can break 1000 IOPS with certain access patterns.

The second graph on this page [tomshardware.com] illustrates the extreme IOPS advantage of Flash for certain applications. Disks are limited by head actuation and rotation latency. This is why enterprise storage vendors have been pursuing [theregister.co.uk] Flash aggressively [forbes.com]. That's what this story is all about.

The dream is to host the same IOPS in with an order of magnitude less physical space, power, heat, etc. If you don't need thousands of IOPS (and most PC users don't) then it isn't very interesting. If you happen to run an OLTP system with thousands of reads/write per second it means a great deal.

Shouldn't that be...

[vjoel]

IOOPS?

The Name Implies

[Chillintau]

With a name like Quicksilver, it must not be RoHS compliant.

Re:

[thedrx]

I wish I had mod points right now, most mods are probably gonna miss this.

Re:

[account_deleted]

Comment removed based on user account deletion

oooooooh

[amnezick]

with 1M IOPS she won't even know how many times she's done ...

Does this mean...

[Hurricane78]

...that they will be worn out in 0.1 seconds? (If typical wear-out numbers apply.)

I'll pass, and rather go with something reliable... ...now where did I put my chisel?

Slow

[matt_martin]

meh ... MRAM runs faster than 25 million ops/second !
http://everspin.com/ [everspin.com]

Yeah, OK , I work there - std disclaimers apply.

Third I/O Broke This Record a Long Time Ago

[Ken Randall]

I personally have seen their device running standard benchmarks on shipping hardware at over 1 million IOPS. www.thirdio.com Also, a company called Violin Memory claims to have broken the 1 million IOPS barrier recently as well. As IBM's technology is at least a year away, I'd have to say I'm more impressed with faster solutions that are already available.