the storage anarchist

1.026: development strategies for solid state storage

There have been some good discussion started on last week's flash wars post. This week, Marc Farley has extended the conversation a bit to include the perspective of NAND-vs.-SDRAM for I/O caching and raises some of the challenges of using solid-state storage merely as a disk drive. I've commented on his post with some added perspective, and I encourage others to weigh in with their own points of view, both here and over on Marc's blog.

Another battle front in these so-called flash wars is whether or not application re-architecture and custom development will be required to leverage the value of NAND flash.

Clearly, integrating NAND as a cache buffer in front of spinning disks might require some hefty integration work - or perhaps not, as has been suggested by the ZFS folks. On the other hand, using a flash SSD in place of a bunch of 15K rpm disk drives requires little more than segmenting the application I/O workload onto the faster media - database administrators and applications developers already do this today when they segment tables and indexes between 15K rpm and 10K rpm drives (for example).

My own opinion is that while solid-state storage will enable application architects as programmers some new opportunities, the fact is that most applications today are able to gain incredible acceleration from DRAM-based solid-state storage and from intelligent cached disk arrays with little or no programming required. In one test case I've seen, simply moving the LUNs off of the 8 heaviest utilized drives from a workload that spans 192 15K rpm 146GB disk drives onto 8 146GB enterprise flash drives resulted in reducing average response times across all the drives by more than 60%. That's a huge improvement that can drive significant ROI without the inherent overhead or complexity of the development/test/change control process.

Not to be outdone by Marc's near-daily vidblog posts, I was recently interviewed on the subject of developing for flash technology by the EMC Developer's Network. If you are an applications developer interested in what it takes to leverage the performance benefit of flash, you might enjoy the interview.

And if you develop applications or tools for use with (or on) EMC products who isn't really interested in developing for flash storage, you might still want to do some social networking with other developers at the new EMC Developer Network - it's the essential community for the EMC developer.

 

technorati tags: enterprise flash drives,EFDs,EMC,Symmetrix,NAND flash,EMC Developer Network,storage cache,application acceleration,database acceleration,SSDs,solid-state storage

1.025: flash wars and the great debate

No, I'm not talking about the on-again US presidential debates scheduled for tonight in at Ole Miss.

Nor the decades-long debate over whether Nikon or Canon make the best intelligent TTL flashes for their cameras. (I choose Canon).

I'm referring to the ongoing debate about where NAND Flash should be used - server, network or storage.

It seems that everywhere I turn I find proponents arguing about where the technology is "best utilized." And like the presidential debates, the positions tend to fall along party lines: server vendors and their suppliers insist you can't get everything NAND has to offer unless you install in right next to the server CPU, ideally with a native memory bus or I/O bus (e.g. PCI/x) interface; the external storage vendors and their supply chain insist that the disk drive form factor and interface is best; and of course the network storage appliance world insists that NAND belongs as a service in the middle.

And they're all correct.

That's right - while EMC and I are both proponents of using persistent solid state storage as a faster alternative to spinning rust, I personally believe that there is no reason the technology won't be applied at every layer - and probably in multiple places along the same I/O path!

In support of my postulate, look no further than all the rest of computing technology - CPU's, DDR SDRAM, spinning disks, PCI/x, protocols, interfaces, connectivity, etc...every single component - both hardware and software - is today applied in servers, and in networks, and in external storage. We put DDR SDRAM and/or processing power into external storage devices and connect servers to them using Fibre Channel, and we also build servers with terabytes of addressable local memory - both approaches can accelerate existing applications without requiring applications to be re-architected or redesigned.

Is there any reason to believe that NAND (or any solid state storage technology) should be applied any differently?

I think not.

So why the continuing debate?

 

1.024: something you should know (about xiv)

Here in the States there is a radio show hosted by Mike Carruthers called "Something You Should Know." Each weekday on the show, which is broadcast by 150 radio stations across the most of mainland US, Mike interviews interesting people who have information about something you should know.

I think he needs to interview an IBM "Top Gun" customer service engineer soon.

Why? Because they know something that YOU should know -  something that IBM marketing, bloggers and your IBM XIV salesperson probably have been neglecting to discuss honestly with you.

It's the answer to this simple question:

"What happens if a second drive fails before my XIV array has completed rebuilding the first failed drive?"

Now, I'm sure that you've been told that the XIV can rebuild the data on a failed 1TB SATA drive in something like 20-30 minutes. And you probably understand that this makes for a very short window of opportunity that a second drive might fail on its own.

But the probability isn't zero - it can't be. Especially when you factor in human error (wrong drive pulled) or adjacent failures (a node dies). But let's not argue the math - let's just explore the results of such a double failure for the moment, irrespective of the probability.

And lest I be accused of spreading FUD, I won't even tell you the answer (until after the break).

Go call your IBM customer service engineer and ask him or her the question (I'd suggest not asking your sales rep - at least, not if you want an honest and complete answer). Note that you may have to ask specifically to speak with a "Top Gun" storage CSE - not all of IBM's service engineers have been trained on XIV service yet. But the Top Guns have.

You might want to be seated for the answer...the chairs are still available.

 

1.023: it's just a flash-y science experiment

And now, my oft-requested take on the 1 Meellyun IOPS flash technology science experiment that IBM is promoting so heavily:

Way Cool.

That's right - Barry Whyte and IBM's Almaden Lab team are to be congratulated for their accomplishment, as I actually did in the first comment to BarryW's boastful blog post on the event. This is indeed an important milestone on the road to wide-scale commercialization of solid-state persistent storage, even if it isn't an actual product announcement (IBM admits you can't buy their experimental configuration for at least 9-12 months).

Commendations all around...

But surely you don't think that's all I have to say now, do you...

1.022: are you ready for IBM's frankenstorage announcement?

I'm not going to Montpelier next week for the Big Launch. For some reason, IBM seems to have forgotten to send me  an invitation. Apparently as much as the storage press corps and the blogosphere have come to expect me to do IBM's marketing for them over the past few weeks, I've not yet earned the early access privileges offered to the "real" press.

Oh well...I'll cope!

I gotta believe this is going to be one weird announcement, though, with lots of positioning and posturing to get all out in one event. I'm predicting the announcement will include:

• the new XIV, which is supposedly going to obsolete every single storage platform on the planet (except any that IBM sells, of course),
• a raft of ho-hum catch-up enhancements to now elderly ancient DS8000
• the new mid-tier DS5000
• the SVC-with-flash Science Experiment

Not to mention the expected release of the first IBM-branded Diligent-based VTL and probably some backup de-duplication capabilities for Tivoli Storage Manager.

That sure is one helluva smorgasbord of unrelated (if not self-competitive) stuff to include in a single announcement.

One thing's for sure, though, the attendees are going to be armed with lots of questions, if coverage and readership of my posts on the stealth launch of XIV and its absolute lack of Green-ness are any indication. Not to mention Chuck Hollis' outing of the DS5000.

Should be an interesting couple of days in the foothills of France...

 

1.021: bass-ackwards marketing

A coworker sent me this link a while back with a suggestion that I might want to use it in my blog. At the time, I really had no idea what to do with it, but now I think I do.

First, the link. Go ahead - click it, then come back.

That's a prime example of bass-ackwards marketing, if you'll excuse the pun. I mean really, who'd ever market anything as a "Real P.O.S." - even if it actually IS a POS?

Over the last several weeks, we've been witness to what could just be the biggest bass-ackwards marketing move ever. Who would have thought that IBM would stealth launch the GA of exciting new products like the first IBM-branded XIV Storage System, the brand-new DS5000 Storage System and a whole suite of DS8000 enhancements? And then, amidst the ruckus that Chuck and I seem to have started filling the void created by IBM's marketing department, IBM puts on a full-court press push to promote a science experiment using NAND flash behind an unsupported configuration of SVC nodes - a configuration that they publicly admit won't be available to customers for at least 9-12 months. Complete with IBM blogger air cover from BarryW and from TonyP. Meanwhile both  remain deathly silent about everything else.

Promote the stuff you can't buy yet, and say nothing about the new stuff that you CAN buy.

Seems so...Alice-in-Wonderlandish, doesn't it?

Now, if I'm reading the tea-leaves (and TonyP's intentional hint) correctly, the press is supposed to take the Blue Pill and show up at the System and Storage Networking Technical Symposium on September 8-10 in Montpelier, France to get the scoop on all this, and more. And the press seems to have calmed down once they received their invitations (mine must have gotten lost in the mail).

Curiouser and curiouser, IBM's marketing team has even co-opted "Information Infrastructure" - a term near-synonymous with EMC - for the title of their Symposium, calling it the "New Enterprise Forum: Information Infrastructure."

Thanks for the plug!

UPDATE 4:45pm EDT Sept 3, 2008: Changed above link to search for the possessive complete string "EMC Information Infrastructure" - returns about 25,200 Google hits, while "IBM Information Infrastructure" returns merely 7,120-ish hits.

technorati tags: IBM, XIV, DS8000, DS5000, EMC Information Infrastructure, bass ackwards