Skiing And Storage

OK, this is going to be an off-beat post.  I'm on vacation this week (sort of), skiing in New Hampshire. 

So, what can possibly be the connection between skiing and storage?

It'll be a little strained, but I'll get there ... just be patient.

A Bit Of Skiing Background

I've been skiing semi-seriously for about 5 years.  I have no natural talent for skiing, but if you do anything often enough, you achieve a certain level of proficiency.

Skiing conditions in New England can charitably be described as "variable".  That means that you're not quite sure what Mother Nature is going to surprise you with on any given skiing day.

The snow conditions can range from fresh powder, to groomed packed powder, to ice crust over powder, to frozen golf balls, to blue ice sheets, to gloopy mashed potatoes, to patchy cover with rocks exposed. 

And I try to ski on all of it.

I think the folks who ski in the Sierra, or the Rockies, or the Wasatch range have it too easy.  You really want a terrain challenge?  Try some of the "snow" in my neighborhood.

Early on, I realized that different types of skis made a big difference in how I could ski.  Over the years, I ended up with a selection of different tools for the job: powder skis, ice-carving skis, general-purpose skis, and -- of course -- my rock skis for when the snow is really thin. 

Grab the wrong pair of skis on the wrong day, and -- yeah, you can make it work after a fashion -- but it's not an enjoyable experience, at least for me.

So What's The Connection With Storage?

There are those who promote the idea that, when it comes to storage, it's possible that one size fits all. 

Regardless of your needs or your situation, they claim a single storage product can do it all for you.  Why needlessly complicate your life with multiple storage products?

I just have to shake my head when I hear this sort of thing.

Now, if you're a beginning skier, of course you start with a single pair of skis, don't you?  And, hopefully, they're the soft, forgiving kind that help you learn how to turn (and stop!) with confidence.

But, as soon as you achieve any proficiency, you ditch those learner skis if you can, and step up for a hotter pair.  And then you realize that hot pair of skis really don't do well on powder, or sheets of blue ice, or frozen golf balls, or rocks, and you end up finding ways to eventually build out your ski portfolio.

Is storage the same way?  If you're new to the whole world of network storage, do you start out with a simple, forgiving environment that helps you figure out what you're doing?  And, once you achieve proficiency, do you inevitably step up to more differentiated and specialized tools?

The Magic Morphin' Ski?

Now, thanks to the joys of New England weather, it's often the case that you'll encounter multiple snow conditions on a single run -- here's a bit of groomer, then some powder, and then some blue ice sheets, and finally the frozen golf ball crud at the bottom.  Or a nice rock.

So, like most people in my situation, I try and compromise with the "best fit" for the conditions, but it's a static choice, and no one wants to haul a bunch of skis to the slope.

Now, it's not practical to switch skis when you're in flight, but -- wouldn't it be cool if your skis could detect the snow conditions, and react in a meaningful way?

I don't think that's possible with today's technology (although I bet someone will try!) since the physical characteristics of each ski type is very different. 

Except for the rock skis -- they don't matter ...

Too bad we don't have magic, morphin' storage - yet.

Magic, Morphin' Storage?

Wouldn't it be cool if storage could detect what you're trying to do, and optimize itself in a dramatic and meaninfgul way to respond?  And do it in such a way that it happened in flight, automatically?

Now, if you dig through various vendors' literature, many of them will claim to do something similar to this, but -- when you dig into the gory, boring details -- they're either doing it over a very narrow range, or they aren't controlling all the resources involved, or it's a static decision that requires re-configuration, or other limitation. 

None of this is magic, morphin' storage -- at least, not the way it could be.  Ideally, a very wide range of cost/performance/protection tradeoffs.  All storage resources (cache, processor, disks, bandwidth, etc.) being orchestrated together.  Optimizations happening continually, immediately and transparently.

And, of course, the all-important auto-detect-and-respond function ...

I think the stakes for doing this right -- eventually -- are rising.

If you think about it, there's a lot at stake in optimizing the different kinds of storage media now available.  At one end of the spectrum, we've got 1TB drives -- very cost-effective, but not exactly the performance champs of the industry.  At the other end, we've got enterprise flash drives -- delivering mind-bending performance, but at a substantial cost premium.  And, of course, everything between.

There's also big potential in dynamically optimizing in how the storage is used.  Different protection schemes (RAID 1, RAID 5, RAID 6 et. al.)  Different layouts (stripes, etc.).  Even different ways of using less -- thin provisioning, data deduplication, single instancing and so on.

Not to mention dynamically optmizing array resources (think cache, processor, bandwidth, etc.)

But, in all fairness,  there are big parts that are missing from a near-perfect solution.  I don't think anyone is doing this as well as it could potentially be done.  Not EMC, not HDS, not NetApp, and certainly not HP or IBM ...

It's just too hard for customers to get to the optimized state.  And, even if you do, the dynamic optimzations don't happen fast enough.  And no one wants to spend weeks writing static performance policies that end up not working so well in the real world.

But even if and when the technology is available, people will need time to learn to trust it.  As with most any dynamic optimization technology, most people will want watch what it does for a while before they turn their backs and let it do what it does.

Back To Skiing

If you're a proficient skier, you can't imagine owning only one pair of skis. 

And, I'd offer, if you're a proficient consumer of storage technology, you can't imagine only using a single storage product.

Unless a single technology platform did it all for you, and could do it far better than any collection of individual technologies, and do so transparently and automatically.

I know that if someone built a magic, morphin' ski that could act like the best of all of them, and do so dynamically, without too much thinking on my part, I for one would be very, very intrigued.  Although I'd be very, very careful on my first run ;-)

And I think the same dynamic would play out in the storage world.

Comments

[Dan Murphy]

When I saw the "Skiing" title, I knew I had to take a look. And since I know a lot more about skiing than I do about storage, I'll stick to what I know. Plus, you're on vacation - relax. ;-)

Skis these days are infinitely more versatile than, say, 15 years ago. I used to show up at the mountain with two sets of boards, ski the GS boards in the early AM, then after a few hours, put the slalom boards on. Now, with the new shaped skis, I feel no need at all to do the switch. My current skis (K2 Apache Recons) will make a nice tight turn, and will rip fast GS turns when you want to open things up. But wait, there's more. They have even performed admirably out west in 2-3 feet of freshies, such as we saw 3 weeks ago at Alta (though I do toy with the idea of getting bona fide powder boards for those trips out west).

As for storage, ah, well.... I'll leave that to you, Chuck. May the powder gods be with you.
Dan Murphy

[Chuck Hollis]

Hi Dan

I was skiing mostly on my K2 Apache Strykers this week -- shortest ski I've ever skiied on -- and it was mostly OK.

But, I have to say, it's still a tradeoff.

I'm not getting rid of my Volkls, nor my Atomics, nor my trusty rock skis anytime soon ... ;-)

[Keith Norbie]

Chuck,

To me to are one of the funniest guys in this game. Since skiing and storage are both my area of expertise I had to chime in. It is funny how many parallels there are. We both are fans of the "new house" in both areas. Skiing technology and product appeal have both advanced. I used to run a ski school as well as occasionally raced so I can relate to having "purpose built" gear. Foam fitted boots just seem to fit and feel better than not.

I'm probably most humored by the parallel with the value of guided expertise in skiing and storage. I was an average Joe skier until I had the good fortune of training for a couple years with a PSIA fully certified guy (he was also an examiner and ran a race program at Mt. Hood). I was stunned as to how much I didn't know. Funny how that applies in storage these days. Through his guidance I could leverage "purpose built" or all-in one gear so for me it was as much technique as gear. I also was able to obtain PSIA full certification which means you can teach any skier in any condition AND facilitate change.

These days I'm a little more into snowmobiling (and VMware) but the kids will probably migrate me back into something that mutually races the "AT" (hear rate). As for Storage... well just like trying to get skiers away from overly pressure/edging their turns and instead focusing on "new school" concepts like moving the center of mass downhill I'll stick to the "new house". Storage doesn't seem as fun (or as benefitial) the old way. Keep up the heat within EMC to innovate (along with Chad).

BTW - Volkl/Technica/Marker (just like Ping)end of conversation. (it's also funny to see how well new skis work without a good tune)

[Dan Murphy]

BTW, we'll be anxiously waiting for the "Mountain Biking and Storage" article as soon as the trails clear up.