I'm just fascinated by the fact that in 2016 there are still huge companies like this making their money on ancient tech that is insanely overpriced to the alternatives.
(~20 years ago when I started working in software I thought I would eventually get how these enterprise software giants actually were worth it. The more I learn and age...)
In the end I guess what companies like Dell and EMC do is to provide access to tech that otherwise (because of cultural reasons) is not available to non-tech companies. But.. anyway, all of this seems like it's ripe for disruption.
"I'm just fascinated by the fact that in 2016 there are still huge companies like this making their money on ancient tech that is insanely overpriced to the alternatives."
You're not wrong.
The problem with your point of view (which I share - it is also my point of view) is that you're operating on the wrong layer of abstraction.
These customers don't need a server (or a network or a SAN) - they need business objectives to be met - again, at a very highly abstracted layer.
You're thinking about technology and capabilities and features and configuration options ... and maybe you're even thinking about price ... but that's not the layer they are operating at. They need an airplane to be built. They need a mine to be dug. They don't care about LUNs or stripe sizes or jumbo frames. They do care about contracts and legal commitments and "synergy". And that's why they rely on heuristics like business lunches and enterpris-ey sales guys and CIA guys giving keynotes at RSA.
I don't like working at that layer of abstraction so I built a business that operates on the layer I like to work at.[1] But as much as I dislike the higher layers of abstraction, it would be childish of me to think they were "wrong".
In some cases it's because stability and repeatability are more important in many aspects. At the level where enterprises are buying components to interface directly with people on a regular basis (e.g. workstations), having a slightly more powerful workstation (20%) generally does not compute into a person that is 20% more productive, but having a workstation that fails to function 20% of the time can definitely lead to a near 20% loss in productivity. Large enterprise providers leverage knowledge of failure rates over vast quantities of almost identical (comparatively very variances, at least) computers to ruthlessly identify any weak links in all parts of the computer, because their service contracts often make them responsible for these problems. This goes as far as creating their own mainboards and BIOS, RAID firmware, etc. Enterprises take advantage of this for both their initial good stability rate and ease of replacement when there are problems.
To put this in another perspective, the vast majority of people buy factory model cars, maybe with the occasional dealer add-on. People could buy custom made cars, and if it was common enough, they might actually achieve a cost lower than MSRP of what Ford, Honda, Toyota, etc sell, by slapping together the freely available parts (from an industry which would look more like the PC parts industry). The average quality would be a big question though, and I'm not sure a lot of people would be willing to deal with that for their workhorse vehicles, that need to be relied on to get them from point A to point B reliably.
What's interesting is your auto comparison... At around $80k, having something custom built/rebuilt starts to become an option... At over $150k it's absolutely an option.
Though the majority of cars on the road have a street value of under $25k, and/or are under financing. Getting financing for a one-off isn't really possible unless you don't actually need the financing.
At a large level I think it's a matter of how costly to you is if it fails. For an enterprise, there can be chained ramifications to problems (low-level worker has problem finishing some task so manager is delayed in getting some info so division head is delayed in some report and CEO has to wait another day before he can correctly assess a company wide status report which may delay action until an opportunity is lost, etc), as well as an overall reduction in productivity due to problems when hardware or software that serves a lot of employees is problematic. For the average person, a car is a big investment. In both cases the stakes are high (but in different ways). For a single person buying a computer (when so often data is backed up or remotely available and another computer can be used), or even an enterprise buying a company car for an executive, the cost of that component having problems is not nearly as high.
At the point where the failure of the item costs little but the interest in the item is high, it can make sense to go a more custom route.
I've heard the "insanely overpriced" comment before - compared to what? The open source/cheaper alternatives tend to be, bluntly, a joke when it comes to the data storage market. They all either have glaring availability or feature gaps that people who champion them try to paint as "not a big deal". It's not a big deal until your fortune 500 relies on it.
You don't buy them for when they work. You buy them for when it goes wrong. At that point EMC type firms will throw engineers at the problem and provide plenty of relationship managers to help you keep your job (sorry, assist in ensuring that your business relationship continues to be mutually beneficial).
I've been on hours-long support calls with EMC and they acted like every other vendor in the world blaming your month-old firmware and deny deny deny and then "oh my shift is up", so whatever works for you if you want to throw money away.
It comes down to fixed resources and reducing operational costs. Yes, you could write your own iSCSI driver, running against a large distributed drive system, or other interfaces, but people don't. There's also a significant cost difference in manufacturing a run of 10k system boards vs 1M of them. Tooling alone is very expensive, and retooling more so.
EMC in particular is fairly custom hardware and software and still tends to deliver better value and consistency than the competition. EMC offers NAS options that has always been ahead of its' competitors. I say this as a passive observer as I've been in software development more than anything operational for the better part of two decades, but I remember in the mid-late 90's that what EMC offered simply had no competition at that time.
Today, it's entirely possible to build a software system similar to what S3 does in-house, but that isn't always what a given system/application, especially legacy systems, require. Not to mention a black box that's mostly closed vs needing specialized software management in-house. There are operational costs to this.
Using a single EMC device vs. hiring 3-5 FTEs making > 80k (120-150K after employer-side taxes) for 5+ years? That's a few million right there. This is often the trade off that EMC offers.
Dell itself offers consistency in broader terms... the same/compatible hardware across all employees... all the same docking stations, all the same monitors, etc. There's value in that as well.
If someone buys a big EMC system, they aren't just paying for hardware and software. They're paying for EMC to also integrate it into their other systems, set it up, train them to manage it, provide custom coding if their environment requires some unusual change, give early access to fixes for problems they experience, etc. There's a support contract with SLA. There's free overnight shipping of components, and field techs that'll fix the hardware for the customer.
Can someone else do it for cheaper? Maybe. Probably. But to a large degree, it's a "no one ever got fired for buying IBM" situation that you'd be fighting against.
Even if it were possible to do cheaper in budgetary terms, it would be way riskier, and that's a serious factor. I think some technologists tend to overestimate the ability of a (non-technology) enterprise to roll their own (without significant risk and overhead) and underestimate the real costs even just from a budgetary perspective. It's apples to oranges.
Lest this be marketing for price is no object, this doesn't mean anything goes. That's where competition is important. And that includes from open source based solutions and smaller players. The key word is "solution." You can bundle up FreeNAS on your own well rested hardware, customized to the customers requirements and couple it with mucho support, hand holding, and SLAs. Oh, and proving financial endurance of thr organization helps too.Then you might at least be comparing something in the same universe.
Fortune 500 doesn't care about saving 20 dollars only to put the other 80 at risk. And rightfully so.
If I have a business making $100M per day using "ancient tech" that costs me, say, $100M per year, I might be interested in switching to the Lambdarific Supracloud, but my interest level will depend a lot more on the risk of failure than on any price difference.
Dell makes most it's money providing enterprise volume orders of desktops/laptop computers, repair parts, and certifying+insuring people to work on these computers.
EMC also contractually guarantees data integrity with some of it's higher end products/contracts. As does IBM/HP/Oracle. All your FOSS tools don't offer this, so your company has to shoulder this responsibility yourself.
Underwriting insurance on multi-million dollar contracts isn't something you can just disrupt unless you already have the funds necessary to burn all the way though series B laying around on your person. In which case, isn't their an easier idea to execute?
>Dell makes most it's money providing enterprise volume orders of desktops/laptop computers
I wish they'd bring back the E6400/E6500 style of enterprise Latitude laptops, but of course with modern internals. These newer ones are much uglier and the screens are too short.
Economies of scale. Everyone wanted a high def 16:9 display, so the dyes for 4:3 became too expensive. I also miss my visually perfect 16:10, but alias time marches on.
I completely disagree. I'm sorry, but I find it impossible to believe that enterprise IT departments and businesses were jumping up and down for 16:9 displays. Maybe all the stupid consumers wanted them, but consumers are not the target customer for enterprise laptops.
I think it's much more likely that the laptop makers saw an opportunity to save money by going to 16:9 because of the massive growth in 16:9 LCD screens thanks to TVs, because the LCD factories were pushing them for economies of scale.
> I'm just fascinated by the fact that in 2016 there are still huge companies like this making their money on ancient tech that is insanely overpriced to the alternatives.
It is basically about switching costs. Even if you have the money you don't want to disrupt your operations because IT is doing a monster migration to different technologies.
I guess what I'm saying is: the companies that are in this situation should hire top-notch developers to solve the problem. But the whole problem is that these companies can't hire top-notch developers because of who they are.
You've just answered your own question. It makes no sense for a company to hire "top-notch developers" to solve "the problem" that storage vendors have already solved (because they have truly top-notch developers with vast experience in this specific domain).
That would turn out to be way more expensive than just paying what you consider exorbitant prices, and way more risky, i.e., it's not like it's easy to find top-notch developers, and even if a company could hire some, it's very unlikely that whatever they developed would compare well, by any metric, including cost, to what the company could get by buying a ready-made solution from EMC, HPE, et al.
I think that's a fairly narrow view of the options. Sometimes you want to focus your company on the markets you are interested in and have cultivated talent to excel at. Not everyone wants to be a software company, or wants to write internal software. Some companies have a division of lawyers, some interface with an external firm. At some point, depending on the work being done, it makes sense to use one over the other. I don't see any difference between that and whether you have your IT department assembling your own hardware and/or developing your own software, or outsourcing that to some other company. It all comes down to how much money it saves, and how many more options it opens up, and how much it increases or decreases operational complexity.
> : the companies that are in this situation should hire top-notch developers to solve the problem.
This is a simplistic view. You can even imagine that all the issues can be solved with free open source development but the organization works in a specific way and changing the way they work, in that scale, involves a lot of risks.
It's probably not ripe for disruption. There's too much furniture to move. And "ancient tech" lives forever because it's a solved problem - one need only develop replacement parts as they go obsolete.
Until you've "done what they asked for" and then have 'em be completely unprepared for the ripple effect, it's hard to explain the phenomenon. The status quo, especially in a low interest rate economy, is a really Big Thing.
Most of the Fortune 500 still use IBM mainframes, which are still the children of the S/360 launched in 1964.
The interesting questions include: how long does it take to move off them; how reliable is the move; and what happens to your quarterly results doing a multi-year transition?
Another one is: How long is the current CEO going to stick around to manage the process?
Syncplicity basically lets companies have a Dropbox-like product that's run on EMC hardware inside of a company's data center. There's a clear migration path to a 100% hosted solution when a company is ready to let go.
You'd be surprised how many companies consider it in their best interest to run their own servers.
It is solid state will be the first place that EMC has faltered. And it has NOTHING to offer. XtremIO doesn't count because it is going to be deprecated soon and still is not a part of vBlock so its future is sketchy.
Eh? XtremIO is selling like bananas, they're like 40%+ of the all flash market. (I don't work for EMC but I do work in the DellTech federation and talk to people).
Thing is, much of that stuff is transitioning to software-defined storage eventually. But that's what VSAN and ScaleIO are about.
(~20 years ago when I started working in software I thought I would eventually get how these enterprise software giants actually were worth it. The more I learn and age...)
In the end I guess what companies like Dell and EMC do is to provide access to tech that otherwise (because of cultural reasons) is not available to non-tech companies. But.. anyway, all of this seems like it's ripe for disruption.