From the earliest days of MongoDB, co-founder Eliot Horowitz planned to build a managed database service. As he stressed in an interview, Horowitz knew that developers wouldn’t want to manage the database themselves if they could get someone to do it for them, provided they wouldn’t sacrifice safety and reliability in the process. The natural complement to open source, in other words, was cloud.
This isn’t to suggest cloud will kill open source. Though Redmonk analyst James Governor is correct to suggest that where developers are concerned, “Convenience is the killer app,” he’s also right to remind us that open source “is a great way to build software, build trust, and foster community,” factors that cloud services don’t necessarily deliver. Even as enterprise customers embrace more Software as a Service (SaaS) vendors like Snowflake or Datadog, open source software will matter more than ever.
Cloudy with a chance of open source
This fact can be overlooked in our rush to cloudify everything. Donald Fischer, CEO and co-founder of Tidelift, said, “Ten years from now much of the complexity around managing open source will be invisible to developers in much the same ways that cloud computing has made people forget about server blades and routers.” Responding to this sentiment, Hacker One CEO Marten Mickos stressed, “We simply MUST automate and package away the current complexities, because we are already busy creating new ones.”
While this sounds great, not everyone is enthusiastic about the trend.
SEE: Special report: Prepare for serverless computing (free PDF) (TechRepublic)
For one thing, as analyst Lawrence Hecht pointed out, it’s not clear we “want [open source] to be invisible” to the user. Sure, we might want to eliminate the bother of managing the code, he continued, “but having an auditable trail is valuable.” Even for those who don’t want to inspect or compile source code (and, let’s face it, that’s most of us), it’s useful to have that access, even if we outsource the work of digging into it.
In addition, there’s another risk, highlighted by Duane O’Brien: Eliminating user visibility into the open source software that powers managed cloud services “will also have the effect of adding an insulating layer between users and contributors. That insulating layer will further propagate the notion that open source is something done by other people, with several additional side effects.” One of the most deleterious of effects? It potentially exacerbates the sustainability of open source projects, as Alberto Ruiz noted. It may also reduce some of the enthusiasm developers feel for getting involved, Jason Baker argued.
But, really, this isn’t about cloud versus open source. It’s really a matter of shifting the focus for end users of that software, as Fischer went on to stress: “The analogy of cloud computing vs private data centers illustrates the opportunity: specialists doing the generic work upstream, freeing up time and brainpower to focus on new organization-specific capabilities further up the stack.”
Even for companies that offer proprietary services, open source is essential. Snowflake just went public with its proprietary data warehousing service, but underneath it’s open source software like FoundationDB. Datadog is similar, with Elasticsearch under the hood. And so on.
We can be grateful for these SaaS companies that make it easier to consume open source software even as we recognize that they simply couldn’t exist without open source.
Or, as Randy Shoup put it, it comes down to a convenience calculus: “If we have to operate infrastructure, we strongly prefer open source. If we can buy it as a service, we don’t really care what’s inside.” But the reason end users needn’t care is because builders continue to care a great deal about open source. That isn’t going to change anytime soon.
Disclosure: I work for AWS, but the views herein are mine and don’t reflect those of my employer.