Joyent CTO Bryan Cantrill Discusses Docker Containers, Industry Challenges, and the Future of Cloud‑Native Computing

Editor’s note: This article, translated from TechTarget, shares insights on high‑availability architecture and container technology; please credit the ArchNotes public account when reposting.

To stay ahead in the cloud era, Joyent has placed its bets on containers. In this interview, Joyent CTO Bryan Cantrill provides a deep dive into the state of container technology in 2015 and the pressing issues that need to be addressed.

Joyent has been experimenting with OS‑level containers in production for over a decade and was one of the earliest companies to embrace the technology when Docker emerged. By the end of last year, Joyent rebranded itself as a “native container architecture” company, releasing products such as Triton—a container infrastructure for managing Docker containers—and Containerbuddy, which helps containerize traditional services for deployment anywhere.

Joyent is also a founding member of the Cloud Native Computing Foundation, established in July to create reference architectures for cloud‑native applications and containers. Bryan Cantrill serves on the foundation’s technical advisory committee and recently discussed container trends with SearchCloudComputing, sharing his expectations for the next year and the changes he believes are urgently needed.

“With containers there’s more opportunity and arguably more peril because they allow you to change the way you think about the problems.” – Bryan Cantrill, CTO, Joyent Inc.

Q: Containers continue to attract attention in 2015, but many problems remain. What’s your view? Everyone sees containers as the future, which brings massive interest. Unlike the VM revolution, where users could keep the same mental model, containers require a shift in thinking; they offer more opportunities but also new traps.

Q: Some say containers are popular because they resemble virtual machines. That’s partly true, but containers are only as powerful as the environment they run in. Simply having a container in production doesn’t give you service discovery. The real value of containers lies in building large‑scale systems faster and more simply, which also forces us to solve how to migrate existing systems to micro‑services.

Q: How mature is the container market today? How well do people understand the technology? At KubeCon, the question “Is the container space near peak chaos?” was raised. Everyone I spoke with—developers, operators, cloud vendors—believes we haven’t reached peak chaos yet; they still expect growth and even accelerated growth.

Stability expectations are hard to set because the technology is open source and fragmented. The container ecosystem contains many frameworks and design philosophies, leading to both confusion and competing approaches. Distinguishing boundaries between these areas can be difficult.

Q: Container‑as‑a‑Service (CaaS) is a trending offering from many cloud providers. Are you concerned that proprietary implementations will hurt cross‑platform portability? People are skeptical of flashy CaaS solutions. We believe the future lies in native container infrastructure—not running containers inside VMs but on bare‑metal containers. When containers have their own IP addresses, many networking complexities disappear.

I think some CaaS offerings that merely run containers inside VMs are just virtual‑machine experts trying to dabble in containers without truly understanding the economics behind the container revolution.

Q: What changes does the container market need next? The expansion of concepts should slow, allowing us to solidify useful ideas and produce meaningful production‑grade solutions. We still lack clear guidance on many fronts.

You can deploy Triton in production and use Docker Compose for adaptable services, but many related technologies still feel immature. The community should stop bragging about code contributions, download counts, or commit numbers, as these metrics can obscure real progress.

By 2017 we expect to have a firmer grasp of which concepts fit which scenarios, with likely further technical breakthroughs that improve robustness while reducing uncontrolled expansion.

For more high‑availability container articles, see the linked resources at the end of the original piece.