[This article was originally written by Brendan Dillon.]
We’ve had a few posts recently on Docker (see here, here) and CoreOS (and here) and generally we’ve focused on the details of how you can easily get your Docker logs sent up to Logentries. But recently I’ve been thinking beyond just how to log with Docker, and more about Docker Inc. and what comes next. So as we all watch and wait to see what happens, let’s go back to the last great disruptive idea in the virtualization space to see what we can learn…
In 2012, when EMC acquired Nicira, the world sat up and took notice of Software Defined Networking (SDN) – the virtualization of the network. A $1 billion valuation took care of that. But it was the waves of confusion that followed that was most interesting. A simple concept — adding APIs to make networks programmable — was now hugely valuable even though up to that point only a few people had grasped the possibilities and the implications. As the valuation sunk in, industry analysts were falling over themselves to explain how this was the death of Cisco, Juniper, and others. Cisco may have had their own internal wave of panic and confusion, but they came out fighting with a very convincing reveal on how networking, policy and big data analytics would interact and the challenges of virtualizing routers.
Small, seemingly inconsequential ideas can shake industries to their core. Before SDN we saw this with platform virtualization. We also saw how a simple concept like Map/Reduce shook up the database / analytics industry. And right now we are starting to see this with Docker – though this time it looks like the hype cycle is peaking a little earlier. With all of these disruptive ideas the attention tends to focus on their impact on cloud architectures but it is their potential to massively disrupt enterprise computing that is most interesting (and lucrative, for that matter).
Initially the impact in the enterprise context will focus on software build and test environments, where the alignment of skills and need is most obvious. But the promise of portability opens up new service deployment models where enterprise workloads can migrate transparently from laptop to private cloud to public clouds, with the potential for significant performance improvements over hypervisors (and with a loss less baggage to carry).
Looking back over Nicira’s funding in Crunchbase, we can see that they did an $11M B round in Sept 2010, followed by a $26M C round in Feb 2011. So well before July 2012, when the $1B offer came in, the insiders knew that this had massive potential. We can also see that Docker just closed a $15M B round. The 36% higher B round sounds promising (and is fairly modest given the hype around Docker) – but arguably this doesn’t look so hot when you factor in the 90% rise in the NASDAQ Composite since mid-2012. Does this imply that the private market doesn’t see Docker having as big an impact on the enterprise as SDN? It certainly seems to reflect the fact that while Docker brings some very clever management and deployment services around Linux Containers (LXC) – unless it can out-innovate others, its moat (in Warren Buffett’s terminology) may not provide enough protection.
But Docker is no doubt a big deal – once the hype cycle plays out. Portability, reliability, testability all make it an attractive technology for Logentries and our customers. And the news from IBM that Docker can have a massive impact on Hadoop performance makes us think long and hard about what future cloud architecture is starting to look like.
So as the future architectures of massive-scale, distributed systems become clearer, let us finish where we started, with a little idle speculation on what happens next for Docker Inc. Of the big players with deep pockets who needs it the most? I don’t have the answer but I am watching eagerly to see how it unfolds. In the meantime, here at Logentries Engineering we have some experimenting with massive-scale, distributed systems using Docker to be getting on with.