While walking along the beach, a developer stumbles across a magic lamp. Out comes a genie, offering him one wish. Ever-savvy, the developer smiles and wishes for a million more wishes. “Can’t do it,” replies the genie, “I can only grant one wish.” The developer is crestfallen for a moment before the genie continues. “But, if you want the next-best thing, you should check out Spinnaker.”
We at Armory believe that there are four truths that will define an enterprise’s ability to thrive over the next decade. One of these truths is that cloud infrastructure is increasingly fragmented. With a number of competing cloud infrastructure vendors, each constantly releasing new platforms and standards, the smart bet is not to go all-in on today’s best option, but to make a bet on the ability to make future bets. Allow me to explain.
Historically (and currently, for some organizations), getting code into production meant pushing it out to a company’s data center. These paths to production had to address issues around environments (dev, staging, production, etc.), business rules (unit and integration testing, QA, manual review, etc.), and a variety of specific compliance and security policies. The process involved a complex, brittle, hand-scripted path to production likely built on Chef, Puppet, or Ansible, often with little documentation.
These same business requirements exist for enterprises today, except that now workloads are moving out of the data center and into the cloud. Using the same “spit polish and tape” processes to create new paths to production for AWS, Azure, GCP, PCF and Oracle Cloud at the same time is simply unworkable. But even if you commit to AWS, you still have choices around EC2, ECS, EKS, Lambda, and Fargate. What happens when (not if) AWS rolls out something new, but your pipeline only works for ECS? The telcos bet heavily on OpenStack only to have to painfully unwind that bet now that OpenStack is being supplanted by Kubernetes. But what comes after Kubernetes?
Today, enterprises want to leverage the optimal target for the optimal workload, especially as monoliths are broken down into microservices. You may want one workload to go to the GDPR-compliant cloud, one to the cheapest cloud, and one to your data center running Kubernetes. How can you do all this with safety and velocity?
Netflix looked at this problem and created an open source project called Spinnaker, now also backed by Google, Microsoft, and a number of other industry leaders. Spinnaker abstracts out the idea of production away from any specific targets, allowing your engineers to define a single, golden path to production with a standard set of policies, environments, and business rules. Companies gain enormous transparency and flexibility in their pipelines without having to build separate deployment paths to each specific target. Instead, Spinnaker’s 6,000 member (and growing!) open source community makes over 100 commits a day to ensure that Spinnaker can deploy seamlessly across all public and private cloud environments. And when the next big thing happens in cloud infrastructure, such as serverless computing, companies using Spinnaker will be able to quickly and easily take advantage of these innovations without rebuilding their entire deployment pipelines.
Armory is Spinnaker at Enterprise Scale. Armory adds enterprise-grade features, functionality and support to help businesses bring Spinnaker into their organizations and ship better software, faster. If you’d like to learn more about Spinnaker, or how Armory can help you run Spinnaker at Enterprise Scale, reach out to us at email@example.com!