The Lure Of Kit Cars

OK, time for a "guy moment" here.  

All the adult males reading this, be honest -- is it true that at one time in your life you seriously considered building a kit car?

I know I did.  I looked at the sleek shapes, the ability to pick and choose components -- even the potential of converting a somewhat boring ride into a potential thrill machine.

And then reality started to encroach on my fantasy.

I rode in a few, and noticed that there were always a few "details" the owner hadn't quite sorted out yet -- like brakes that worked reliably.  Or a ride that wouldn't jar your fillings loose if you hit a pot hole.  Or starting reliably, for that matter.

Not only that, as I looked at some of the plans, all them assumed hundreds of hours (or more!) of assembly time.  Not only that, a place to do the assembly and a bunch of tools and equipment I didn't own.  

I also had serious doubts as to my own mechanical abilities and commitment.  My initial investment could end up being a half-finished project sitting in my garage for years at a time.  My wife saw that future scenario clearly, and made her perspectives very clear to me in that unambiguous way that she does :-)

Finally, I realized that the whole kit car thing was more of a hobby than an alternative method to procure reliable and stylish transportation.

I passed -- as many people do.

Why Am I Talking About Kit Cars?

Because the current pre-integrated infrastructure vs. do-it-yourself infrastructure debate is splitting along similar lines.

On one side, we have the traditionalists.  If they're in an IT role, they truly believe that they can build something better than rolls off a modern assembly line, backed by hundreds of millions of dollars of R+D.  Better -- and cheaper as well.

The traditionalists are joined by those who are seriously threatened by this new evolutions in the IT infrastructure business.  

Maybe they're specialized system integrators whose skills are not needed when the decision is made to go integrated vs. components.  

Or, more likely, they're one of the component vendors who can only offer up a sketchy blueprint of what a car might look like -- of course, built with their components.

"Avoid lock in!" they'll threaten.  "Stay with best of breed" they'll proclaim.  "Preserve your legacy investment" they'll entice.  Or perhaps go to other marketing words like "open" or "elastic" or something similar.

None of these claims really stands up under serious intellectual scrutiny, but it's all they seem to have -- sound bites.

Driving A Car Vs. Building A Car

Now, make no mistake about it -- the inner geek in me still longs for a scenario where I get to pick the automotive components of my choosing, and have the time and skills to create the car of my dreams.  I get that.  But I now see that as a hobby, rather than a great way to get a car that meets my more pragmatic needs.

And, if you go way back in automotive history (from very late 19th century to early 20th), the only way was to essentially build one yourself -- or have someone do it for you.

But times have moved on in the automotive business, haven't they?

Now, imagine for moment that you were in charge of running a large-ish fleet of automobiles, and you were currently using a kit-car approach to provide the cars for others to use as part of the service you deliver.

You'd be spending a lot of time and effort with your car assembly people.  Maybe lots of debates as to who has the best engine, transmission, seats, etc.  I can easily imagine a flock of vendors descending on you with glossy powerpoints and blueprints trying to convince you that their components are better choice for your next kit car.

Sure, you could standardize your bill-of-materials (and your process) -- but are you really adding value?

Meanwhile, your users of your car service are starting to complain -- and loudly.  The cars you provide them aren't as good -- comfortable, reliable, safe, cost-effective, etc. -- as the ones that other car services are providing.

You might try to argue back with "well, we're different" or perhaps "we build better cars, here's how" or even the more desperate "we've always done it that way".  Over time, none of those arguments will likely stand up.  Your car service customers will figure out that there's a better deal to be had, and that will be that.  

Capitalism at work.

At some point, you -- as the senior executive of the car service -- will turn to your people and tell them that times have changed.  The blueprint + bill-of-materials + lots of effort approach to building cars has passed, and it's time to move on.

And -- technology aside -- that's the exact scenario I see played out many times each week, as EMC and VCE put a Vblock in front of a decent-sized IT infrastructure organization, and softly suggest that times may have changed.

Comments

[Chris Mellor]

Chuck,
Great post; but does it go far enough? Wouldn't it be better to buy from a single integrated vendor rather than from a master assembler who gets the engine and gearbox from GM, the suspension, wheels and floorplan from Ford, and the body from VMcarware Inc? Won't an integrated IT stack system coming from an single integrated vendor, rather than a coalition of three vendors, have even greater integration and be more efficient and perform better?
Chris Mellor,
The Register

[Chuck Hollis]

Hi Chris

Always a thoughtful response, so thanks.

Analogies with automobiles and the car industry will get you only so far, but consider this: almost every automotive manufacturer subcontracts to specialists for key sub-assemblies.

The car vendor (and dealer) is responsible for the end user experience -- and has to figure out which bits of the value proposition they provide -- and which ones are better done than others.

Sure, we could debate exactly where the optimal lines of demarcation might be -- but I'm arguing a different premise: the traditional boundaries are shifting, and very fast.

Thanks for comment

-- Chuck

[BrandonJRiley]

I don't agree that it would be better to buy a car entirely built by one company. BMW uses ZF transmissions. Because ZF makes a fantastic transmission.

As a customer, why do I want BMW to make their own transmissions when they can buy better quality ones cheaper from ZF? I don't.

This whole scenario explains why I am not a huge vBlock fan. I don't believe that an EMC Clariion is the perfect array for every customer, and every application. I don't believe UCS is the perfect server platform for every application. I want the flexibility to choose something else if it will perform better for my particular application.

[Chris M Evans]

Chuck

I think this is rather a forced analogy. Firstly most of the manufacturers of kit cars don't offer total flexibility; it's not like I can take the car you pictured and put a Bugatti Veyron engine in it. Kit car builders have to work in the parameters of what's offered. Building any car is a compromise to what will work mechanically.

Extending to your fleet analogy, I'd ask how many people who get provided with a fleet car (company car in the UK) are actually happy with the car they receive? I think you'll find, given the choice they will prefer to choose something better, but they don't pay for the fleet car, so choose to keep quiet and accept the free transport. There's also the rise in the UK of people opting out of the fleet deal and just taking the money to fund their own (better choice) car.

Sure there will be plenty of customers who like the "single throat to choke" approach, especially if they're provided with the integrated software. But there will also be many customers who prefer to manage their own infrastructure choices for a variety of reasons. Neither route is best, but rather best for that customer in question.

The message from EMC at the moment seems to be that unless you're taking the integrated stack then you're on the wrong path. Integrated is good; non-integrated is bad. Am I reading that wrong? If so, please put me right and I apologise in advance. If not, then I think you're doing a disservice to many customers who just prefer to manage their own infrastructure to suit them.

[Chuck Hollis]

Hi Chris

I think all analogies are forced, so we're in agreement in that regard.

Regarding your key question -- fortunately, the world is not a black and white place, is it?

Our argument is that there are many situations where IT is not adding value by trying to act as the internal system integrator for requirements that are basically the same across a wide swath of IT requirements.

Not only that, but the differences in opex between the folks that are doing it the traditional way -- and those that are doing things the new way -- are stunning, to say the least. They're not only saving money, they're delivering better IT services, period.

To do this, the operational model (and associated roles and responsibilities) has to change as well. Most definitely not IT as usual. Put differently, you can't run a Vblock like it's 2004 :-)

Are their certain IT organizations that ought to be doing things themselves? Certainly -- but not many.

Are their unique requirements that don't lend themselves to standardized infrastructure? Of course -- I see them all the time.

That being said, you'd be amazed at how many IT organizations are essentially re-inventing the wheel -- or the car, using this analogy.

Thanks for the comment

-- Chuck

[VMTyler]

@Brandon-

What percentage of BMW owners know it's a ZF transmission? They aren't buying the best engine and best drivetrain, they're buying the BMW driving 'experience.'

That's the fundamental problem I used to struggle with as an engineer/architect- Your internal customers requirements are 'support xyz apps requirements for processing, memory, storage, data protection.' Period. Not 'build the most technically advanced infrastructure solution.' If a Vblock is sized appropriately to have sufficient compute, network, and storage iops/space, why does it matter if its a Clariion, Symmetrix, or Celerra?

The answer is the same as the BMW example- only us gearheads know and care.

Look at Apple- I bet you at least 50% of MacBook owners don't know what the CPU is in it and probably 98% don't know if they're running a 64bit or 32bit OS and what the advantages of each are. Apple is selling an experience, not a combination of hardware components. It 'gets the job done easily' for so many people, you can tell them about the video card in your Alienware desktop until you're blue in the face. They'll say "that's fine- this just works for me."

That's what Vblock is all about- if you want it to 'just work' get a Vblock. If you want to roll your own, go ahead. EMC has great non-Vblock solutions in a virtualized environment.

Tyler
(EMC Employee)

[BrandonJRiley]

Excellent points Tyler.