Virtualization remains the rock on which all advanced enterprise infrastructure architectures are built these days, so it is somewhat surprising that so little attention is being paid to what could be a truly radical revamping of the virtual layer.
A company called Docker is pushing a new containerized approach to basic server virtualization that some say could drive new levels of performance and system flexibility. It works by allowing multiple applications to run on a single host, rather than filling bare metal servers with multiple hosts that each hold a single app. Essentially, it virtualizes the operating system as opposed to the server itself in order to allow greater agility when it comes to driving resource utilization and bridging cloud-based and on-premises infrastructure. At the same time, it gives dev/ops teams a break, because they can code without caring where or how the app is deployed.
Does this make Docker a better solution for the enterprise? It depends on the application, but in general, there are some key performance benefits, according to researchers at IBM. Using the Linpack metric, the team recently pitted Docker against ESX and various KVM solutions and found that Docker achieved the closest to bare-metal performance of them all, outshining KVM by nearly 50 percent. With only one operating system to accommodate multiple apps, Docker has much less overhead than a traditional VM approach, although researchers were quick to note that Docker is not flawless: The use of Network Address Translation (NAT), for instance, can hamper performance in high-packet rate data flows.
Naturally, this should be cause for concern for VMware, but so far they are showing no sign of it. In fact, it appears that the company has already moved beyond mere virtualization and is now focusing on issues higher up the stack. All of the top releases at VMworld this week have to do with virtual networking and the software-defined data center (SDDC), with integration deals involving top-tier enterprise vendors like Dell and HP at front and center. This could be very good or very bad for VMware in that enterprises that are building their next-generation data architectures are unlikely to mess with the underpinnings of those architectures, but if someone like Docker can demonstrate a better way to build and manage virtual data resources, then the rug could come out from under VMware in relatively short order.
And the fact is that most enterprises are still working their way through rudimentary virtualization techniques like resource consolidation and basic load shifting. When it comes to sharing virtualized environments across the cloud or spinning up new dev/ops instances at the drop of a hat, many organizations simply lack the institutional knowledge to support this level of functionality even if they have access to the technology. And there is still strong resistance to porting mission-critical apps to virtual infrastructure, which could put the brakes on full SDDC deployments long enough for rival virtualization solutions like Docker to work their way into the enterprise.
None of this is meant to suggest that the enterprise industry is about to make a left turn on the virtual layer in order to revamp everything that sits above it. But it cannot be denied that such a move is much easier to pull off in software than in hardware. As the basic server attests, the data industry is not above shelving tried and true technologies like the mainframe if a more effective, efficient solution comes along.
Docker has yet to prove that it is capable of disruption at this level, but nothing yet suggests that it isn’t.
Arthur Cole writes about infrastructure for IT Business Edge. Cole has been covering the high-tech media and computing industries for more than 20 years, having served as editor of TV Technology, Video Technology News, Internet News and Multimedia Weekly. His contributions have appeared in Communications Today and Enterprise Networking Planet and as web content for numerous high-tech clients like TwinStrata, Carpathia and NetMagic.