As an industry, we've spent decades collectively wringing our hands over Evil Vendor Lock-In. And for equally as long, we've happily shoveled hundreds of billions of dollars into these same vendors whose lock-in we allegedly fear: Microsoft, Oracle, and now Amazon. If it weren't for proprietary licensing, cloud hosting, or other evil machinations, we'd live in a paradise of perfect competition and everyday low prices.
That's complete and utter rubbish.
In a sign that our lock-in fears have reached peak paranoia, however, enterprises increasingly eschew commercial versions of popular open source software due to (you guessed it!) Evil Vendor Lock-in.
The problem with this argument, as analyst Curt Monash highlights, is that it overestimates the risk of lock-in while simultaneously underestimating the value a commercial relationship with a vendor can provide.
Lock-in is universal and inescapable
Let's first get one thing straight: There is no such thing as a lock-in-free existence, at least not for an enterprise. The very nature of enterprise IT adoption is such that a decision to use one technology is necessarily a long-term commitment. Enterprises aren't in the habit of going into production on Software A one day and ripping and replacing it the next with Software B. It doesn't happen.
Sure, vendors do all sorts of things to make it harder to replace their software, but proprietary licensing is arguably the easiest to escape. I've talked with more enterprises that are "trapped" by advantageous pricing than disadvantageous licensing. That is, if a vendor sells an enterprise an enterprise license agreement, they're going to have less inclination to try alternative vendors because, well, the software is already paid for.
Perversely, the more expensive the ELA, the less inclined the buyer will be to seek alternatives, even when the software clearly isn't working.
For those who bizarrely see the cloud as their key to a lock-in-free existence, think again. Even if you opt for a multicloud approach, whereby you expect to make applications portable across clouds using containers, the reality is that cloud portability is, as Eric Knorr argues, a "pipe dream." Each public cloud has its own nuances, and those nuances make portability ... less portable.
Before you shout, "That's exactly why open source is awesome! No license or cloud lock-in!" it's time for a reality check. Even if your company invests $0.00 in licensing open source software, it invests lots of cash in people to make it productive for the enterprise. Once those people have invested countless hours in deploying their open source baby, they're equally as disinclined to rip it out.
Besides, as Monash notes, those companies that hope to "do as Facebook" and build out on fee-less open source need to remember: "Facebook has more and better engineers than you do." Compounding matters, Gartner finds that 95 percent of cloud failures through 2020 will be the enterprise's own fault, suggesting that even when given controlled infrastructure, our own organizations still have plenty of capacity for mistakes.
Reality bites.
Avoiding open source lock-in
Lock-in is particularly fearsome when it comes to a company's data. As such, Monash talks with a wide range of enterprises that tell him they're afraid of Oracle database lock-in, "yet they're so afraid of lock-in now that they don't want to pay for ... vendor-supplied versions of open source database technologies" and instead "they prefer to roll their own."
This is dumb.
First, it's dumb because "rolling your own" in no way eliminates lock-in (see above) and introduces all sorts of unanticipated costs (also see above). Second, it's dumb because the relative cost of a Cloudera or DataStax is peanuts compared to incumbent vendors like IBM or Oracle, making the license fee "lock-in" far less onerous.
Third, it's dumb because the value provided far outweighs any lock-in risk. As Monash avows, "The management of even NoSQL DBMS is a big issue, and help in that area has high cash value for customers." This is your data, in other words, and it shouldn't be an area enterprises skimp on.
Beyond databases, however, it generally pays to pay for software, open source or otherwise. Whether you pay or not, however, don't get duped into believing the "no lock-in" myth. You're locked in the second you choose to use any software -- and guess what? Most of the time we're fine with that.