Re: The solid core of the big storage business
"Enterprise IT cares deeply about value per dollar, but have a very different cost for downtime and data loss than SMBs".
Bullshit. Nobody can afford data loss these days. This isn't the 90s. As for downtime; Enterprises can afford to Adobe the world for a day and they're fine. It's a little embarrassing, but banks to don't under because the debit machines stopped working again for the umpteenth time due to bank IT screwup. Target still sells shit, even after screwing up and selling you. Enterprises can afford downtime, data loss and screwups because they are too big for the consequences to truly affect them.
If SMBs or SMEs have downtime they are done. They'll lose clients instantly because the world they live is is ultracompetitive and their customers are fickle. Perhaps more to the point, the large enterprises they compete against will instantly jump all over any outage or data loss and turn a marketing machine worth more than the gross annual revenue of the little guy towards smearing their name and driving them out of business.
Large enterprises can survive government interventions that border on inquisitions. SMBs can be murdered by nothing more than FUD.
"I am subjecting myself to a risk"
And there is my point crystallized: cover-your-ass-a-service. It's what really matters to the large enterprise (and large enterprise admins), not value for dollar.
"They just don't like it when people consider personnel turnover, serviceability, and risk as part of that cost of ownership."
What are smoking, and can you please share? This is exactly what (good, successful) startups want you to consider. The (good, successful) startup scene is about making products that are easy to use and reliable.
They aren't always (I would argue rarely are) cheap...but they generally do quite directly address things like "what happens if people get hit by a bus". Not by making a certification program to ensure that you have a $20k certification path to know how to properly swap a hard drive when RAID fails, but by making sure that you don't need that level of training to make the damned thing go in the first place.
There are plenty of bad startups out there that are little more than con jobs. I have a list as long as my arm...but there are plenty of startups that are capitalizing on the fact that some aspect of IT has become so viciously overcomplicated that it needs a good pruning. That small sector has developed into a "specialization" which now exists only to perpetuate the need for that particular specialization to continue to exist. They then set about automating/software-defining/completely-reinventing-the-basics away the need for that level of complexity in the first place.
Perhaps one of the better examples of this is Tintri. Tintri "just goes fast", and for cheap. There are zero nerd knobs to tweak. There is no "optimizing" to be done. It is faster and cheaper than the alternatives, period.
When and where Tintri gets a win and convinces a customer to buy a box, the second, third and so forth follow in short order. They are doing quite well - and even slowly eeking into the large enterprise - because they made a product that is just flat out better than big-brand arrays it is competing against. And it's better because there are no nerd-knobs to tweak, not in spite of it.
Despite this, Tintri earns hyperbolic vitriol from a great many enterprise admins who've never touched it. It's new and "untested" (bullshit!) they say. They can't get into the guts and tweak it (that's the fucking point!). FUD, FUD, FUD. Cut through it - usually 10 beers later - and they just don't want it around because they either A) haven't used it before so they don't understand it or B) their job wouldn't need to exist if they bought it.
"If any large enterprise could become more competitive and reduce costs by going with storage startups or the cloud, they would have done it and reaped any benefits, which would prompt their competition to do the same or be left holding a higher operating cost ratio."
Bullshit, bullshit, and double bullshit. That's the same sort of "people are rational actors" tripe as American conservatives spew. People aren't rational actors, and corporations/governments sure as hell aren't.
Corporations and governments are made up of individuals. Those individuals - by and large - have little-to-no loyalty to the company or government they serve. Individually and collectively they care about one thing: preserving the money that is their salary.
You don't get a bonus for picking the most efficient product that can do the job within the safety margins. You get a bonus for picking the "known good choice" and never taking any risk, no matter how small. As stated above: companies at the size of a large enterprise are big enough to cope with being massively inefficient, even to the point of experiencing huge outages, data loss and so forth.
Large enterprises don't need to be as efficient as the next guy. They just need to be able to scapegoat someone if something goes wrong. That means that the sole focus of people that work there is never being ion a position to be scapegoated, even if that means it costs the business more. It isn't their money, they just don't care.
"The reality is that we all have little projects we use to test out new stuff,"
Completely untrue. First off, a huge chunk of large enterprises absolutely do not have internal skunkworks for IT. They let their competition do that and they then follow the herd by adhering to "best practices" and the whitepaper farm. Nobody at that level is going to be out of business because their IT plant is a little less efficient than the next guy, and they can skimp on the R&D that way.
"and if anything were so significantly better than the established low-risk vendors we tend to use, they would quickly become "enterprise vendors"."
Nope, sorry. I've seen lots of situations in which the startups pass muster on the technical side, meet all the business requirements and otherwise are well suited to operating in a large enterprise (indeed, already are in some other large enterprises), but are shot down as a vendor for purely political reasons. Massive FUD from an admin or team worried about their own relevance is one frequent item, but by far and away the most common is some pointy-haired boss who just doesn't want to lose junkets or clout with his existing junket provider.
Besides, you act as though SMBs and SMEs just roll up to a startup, swallow whatever tripe they spew and toss complete unknowns into production. That's bullshit of the nth order. 80% of the companies I work devote around 1/3rd of their IT budget to R&D and PoCs. Dev and Test are frequently a substantial size of production, right down to companies with as little as $3M in gross annual revenue.
Now, I'll admit, when you get below $1M GAR, everything changes and IT starts to become largely disharmonious consumer-based pap, but - quite frankly - so is everything at that level. There's a reason a huge % of companies never make it to $1M.
Above that, however, and right up to the point where politics dominates over sheer corporate survival, R&D/prototyping/PoCs are absolutely critical. IN the SMB/SME space you only get one chance, and you're usually toast if you botch it.
Remember also that in the SMB/SME space, IT isn't an empire. We don't get to dictate terms to the business. The business demands we provide a given service - and level of service - and you get what you get for budget; there isn't any more to be had.
In the SMB/SME space, we only keep our jobs if we're more efficient than the next guy, without incurring any additional risk.
In the large enterprise space you only keep your jobs if, when something goes wrong, you can both point to a "best practices" document and claim "everyone, everywhere does it this way" and you can redirect the blame cannon onto another scapegoat.
More than anything, this is what holds back the evolution of enterprise vendors. The "new guy" company is the easy target when something goes wrong, even if it isn't actually responsible. Noone wants to be the one responsible for introducing new technology or vendors to the mix because they don't want to be the scapegoat.
There's nothing there about "fitness for purpose" or "value for dollar". It's just fear and politics. Even if the thing is entirely fit for purpose and provides superior value for dollar uptake will be painfully slow in large enterprise.
There are always exceptions to everything above, but even those - such as Netapp - are very slow growing. Even when offering better for value for dollar and when equally fit for purpose, Netapp will still struggle inside a large enterprise for specific deployments due to individual ass covering and fear...and here's a company that has "made it" and has a "presence"!
People are not rational actors, and collections of people are even less so. Large collections of people are panicky herds, and I find your belief that individuals within the belly of a "whale" will do what's best for the whale (instead of for themselves) quaint and alarming at the same time.
It certainly doesn't align with any of my experience, or my research. Maybe you, personally, are a good admin; loyal and true and working for the good of your employer. If so, that makes you rare...and I hope they buy you a bloody island, because if that's the kind of admin you are, they'd better not lose you.