1996 Flashback: Apache, Linux and the Growth of the Worldwide Web: What this Means for OpenStack and Networking
Sixteen years ago I was leading a team building some of the first web commerce businesses for IBM. We sold it as SaaS called World Avenue and World Distributor (we called it hosted apps) and as a piece of software called net.Commerce. We, the general managers of these new businesses, had to make a decision since we did not have a web server to build our applications on top of. Should we take advantage of a new Linux open source web server available from the Apache foundation or should we hold off until we wrote our own web server? This was such an important moment that Tom Friedman captured a key dialogue I had with my software leader about this decision in his book The World is Flat:
One day I asked the development director who worked for me, ‘Say, Jeff, walk me through the development process for these e-commerce systems. What is the underlying Web server?’ And he says to me, It’s built on top of Apache.’ The first thing I think of is John Wayne. ‘What is Apache?’ I ask. And he says it is a shareware program for Web server technology. He said it was produced for free by a bunch of geeks just working online in some kind of open-source chat room. I was floored. I said, ‘How do you buy it?’ And he says, you download it off a Web site for free.’ And I said, ‘Well, who supports it if something goes wrong?’ And he says, ‘I don’t know-it just works!’
This was my first real general manager role and I was about to put IBM customers on something to which a bunch of distributed developers were contributing. IBM customers! I was intensely scared (and thought seriously crazy), but I knew that, if I waited, the largest IT company in the world would enter the market a year or two later, ceding a giant new opportunities to others uncontested.
When talking about Linux back then, what did people talk about?
- Open source – hah.
- Is it ready for prime time?
- Who will support it?
- How does anyone agree? Herding cats #!@!
You know the rest of this story: Linux became the backbone of the worldwide web and enterprise computing.
Now the world of software is taking over all facets of the IT stack: compute, storage, and network. At Nicira we are pretty confident about OpenStack, but to give the naysayers their due, it’s relatively early days.
But the last 10-15 years has taught me another lesson: the fast eat the slow. And as our backer Marc Andreessen noted, software is eating the world. My big take away from the OpenStack conference was not the image of a software baby swaddled in diapers, it was an emerging giant growing by leaps and bounds. The progress is two short years is astounding. It might not be the only software cloud stack (we have customers that use all of them), but it is moving fast.
The big guys are not asleep here: IBM, AT&T, Red Hat, and HP, just to name a few, are in OpenStack.
Quantum, the networking component, went from a side project to demonstrable working code in less than a year. Some of the largest cloud providers like Rackspace are putting Quantum into production – and it will be a core OpenStack project in the upcoming Folsom release this fall.
The critical component of Quantum is the power to abstract (and then deliver through plug-ins) the rich capabilities of enterprise networking in the cloud. Abstractions matter because developers can build powerful, differentiated services through software abstractions and be shielded from the gorp of the underlying plumbing. This delivers speed to market, speed to service, speed to revenue.
The fast eat the slow. It took Linux just 10 years to become the primary web server in the market (by 2008 Steve Ballmer estimated a 60%-40% split between Linux and Microsoft, respectively). Will it take software 10 more years to consume the rest of IT?
Service providers, enterprises, and government agencies live in the Flat World, a very fast paced marketplace where leadership can shift dramatically. Five years ago, no one thought the iPhone had much of a shot against the Blackberry in the Enterprise. If you are introducing the next tablet, do you want to have the iPad or the RIM Playbook? The fast eat the slow.
When networking (and compute and storage) becomes software, things can happen very quickly. That was not lost on the 1,000+ standing room only attendees of the OpenStack conference. Yeah, people can tell you open source software is scary and imperfect. But being the fast, the first in a market is usually worth a risk.