This is Hyper-Disaggregated

couldn’t resist

I have a lot of thoughts on the path storage is moving which might take a few pages to expound upon. I’ll probably update this as I consolidate my line of thought, but here is what I’m seeing right now:

The “unified interface” jives with the disturbing trend I see from most of the storage vendors. They are marketing to the CFO/CIO/CTO with their pitches of simplicity of use and throwing out the term “IT Generalist” to sell it because they wont have to hire a dedicated storage admin to manage it all. For the C-level that’s a win-win. They can get the latest/greatest and won’t have the ongoing expense of a dedicated worker to manage it.

Those of us in the trenches that have to actually make it work really don’t like that message because I think in our view it devalues the amount of training and expertise we have accomplished and it also is highly inaccurate. So when EMC has their dog and pony show with a kid with an IPAD managing a VNX its a slap in the face to the people who have to make that shit work and its a pretty scummy marketing ploy because the C-level doesn’t know any better (well most of them dont).

Established vendors are still over charging for JBOD and still trying to nickel and dime their customers for features that should be included in nearly all arrays *cough* netapp/emc *cough*

Some major players have tried to buy market share and revenue at the expense of innovation *cough* HP/IBM *cough*

Many of the smaller and new companies are not falling into the trap of trying to be all things to everyone, they are carving out niches and taking away large chunks of business that the major players have taken for granted.

Shoving multiple disparate disk types into an array and auto-tuning it for workload performance is swell, until the workload doesn’t jive with the vanilla auto-tiering schemes and shits itself

De-dupe was much overhyped and has yet to deliver on the promises it has made for the last 5 years or so. I simply don’t find that much value in it yet. Same for Encryption (which should be handled at the client level and not the array anyway)

Through Howard Marks and a discussion on needing new bench mark tools, I saw this link to Demartek digging under the covers of the various IO tools to find that some of those data sets really lend themselves well to dedupe and compression algos more so than others.

SQLIO: all zeros = all worthless, but I do believe there is an option to do random data sets. Though I’m not sure what the output looks like.

IOmeter 2006  = good

IOmeter 2008 = Crap, Why the change?

Full data above at the link. Perhaps my initial take on this is somewhat simplistic, but I would say that the overall issue remains. Simple tools produce simple results.

So this quip from Howard got me thinking about a demo I saw from Nimble the other day:

If that weren’t depressing enough, even the most sophisticated benchmarks write the same, or random, data to create their entire dataset. While disk drives, and most SSDs, perform the same regardless of the data you write to them, the same can’t be said about storage systems that include data reduction technology such as compression or data deduplication. If we test a storage system that does inline deduplication, like the new generation of all solid state systems from Pure Storage, Nimbus Data or Solidfire, and use a benchmark that writes a constant data pattern all the time, the system will end up storing a 100GB test file in just a few megabytes of memory, eliminating pretty much all IO to the back end disk drives, or flash, to deliver literally unreal performance numbers.

During the demo, they run IOmeter on one of their rigs during the entire time. You can see the system chugging along at around 16k IOPS, but without knowing what IOmeter version they are using, that number is pretty worthless. I’m not trying to pick on them, but its something that I was unaware of with IOmeter as well as some of the other tools and it puts the results given into a better perspective for me.

Just an aside, the IOmeter version used in the VMware IO Analyzer tool is version 2006.

Chris Mellor over at El Reg has the specifics. Consider this no big surprise as all the big vendors will follow suit, and the trickle down will hit the storage industry as a whole. The big mystery factor will be at what point will the prices reduce, if ever. Definitely  something to keep an eye on 6 months from now as drive production ramps back up and stabilizes.

According to IDC, Thailand accounted for 40-45% of worldwide HDD production during the first half of 2011. Many factories have been flooded, and in others production has been impacted due to power outages and work stoppages. Approximately half of Thailand’s HDD production capacity has been impacted by the flooding.

The magnitude and duration of the disruption will not be clear until the floodwaters subside, but the industry is already experiencing severe shortages of certain HDD components. In particular, large-capacity SAS and SATA drives are in short supply. HP is working closely with our suppliers to maximize our access to these HDDs. We have a significant advantage in this environment due to our world-class supply chain; however, worldwide output will clearly be much lower than worldwide demand.

This reduction in available supply is causing immediate and significant increases in the prices that HP and all other vendors pay for hard disk drives. Component prices have already increased approximately 20%. In this context HP will be forced to increase the prices that we charge for certain disk drives.

We are hopeful that this crisis will come to an end soon. We are doing everything in our power to gain access to sufficient HDD supply so that we can help you fulfill your requirements. We are also prepared to engage with you to explore alternate solutions where appropriate.

Perhaps putting nearly 50% of the worlds hard drive manufacturing into a flood/tsunami prone lowland country wasn’t the smartest move.

EMC announced that they will be putting a 15% premium for the price of all drives indefinateley because of the flooding in Thailand. That can’t make their customer base very happy. Still nothing says “get that end of the year deal done” like a price jump.

EMC has notified partners and customers that it will raise the list prices of its hard drives by up to 15% beginning Jan. 1 due to shortages caused by Thailand floods. The increases are expected to be temporary, depending on how long it takes damaged hard drive manufacturing plants to recover.

Hot on the heels of Western Digital deciding that they would drop warranties from 3 years to 2, Seagate followed suit (just don’t call it collusion).

WD:

This new warranty policy will be effective for drives shipped from January 2nd, 2012. It is important that you take a moment to update your website(s) and collateral to reflect this change for effected drives shipped after January 1st, 2012.

Seagate:

Effective December 31, 2011, Seagate will be changing its warranty policy from a 5 year to a 3 year warranty period for Nearline drives, 5 years to 1 year for certain Desktop and Notebook Bare Drives, 5 years to 3 years on Barracuda XT and Momentus XT, and from as much as 5 years to 2 years on Consumer Electronics.

Now I can’t necessarily fault EMC for going that route. As a business you can only absorb the increase in costs for a certain period of time before you will have to pass those costs onto the customer. What bothers me more is WD and Seagate dropping warranty support 33%, and in some cases 80%. That’s a cheap move, and it smacks of built in obsolescence for your product line. Nothing gives me a vote of confident like saying “Hey our product is crap and will die a lot earlier than even we expected, so we are going to make sure you have to buy a new drive instead of getting the defective ones replaced”

But lets be honest, both  Seagate and Western Digital have a duopoly on the hard drive business, and they can pretty much set pricing world wide. Lets also face facts that consumer storage is pretty cheap and the margins are continuing to be reduced, but Enterprise class storage is still awfully expensive. I’ve had to make the argument time and time again that the 1TB drive you get over at Frys wouldn’t last 15 minutes in a production SAN environment, many CFO types just cannot grasp that concept and balk when it turns out that the new storage array I’m pushing will cost anywhere from $7000 to $12,000  per TB usable, but thats a rant for another time.