PKS: The Answer for “Everything Else” in your Data Center
Back in 2009 when I worked for a global enterprise, I was part of a major data center consolidation and migration project, which took almost two years to plan, coordinate, and complete (man… I love the enterprise). The scope of the migration included both large client-facing applications, some of which supported 50,000–100,000 logins per day, to much smaller back-office apps and services, many of which ran on a cron-like schedule. The easiest part of the project was to identify and plan the migration for the most prolific client-facing apps. Our team used a ton of in-house domain expertise to determine how to best move these applications — from data replication strategies and networking tricks to bleed over web traffic.
But shortly after we identified game plans for major apps, the harder aspects of the project kicked in. For the remainder of the project, we devoted a majority of our time trying to identify and strategize how to move "everything else" that was running in that data center. From batch jobs and file watchers to automated reports and even packaged software that was used by the NOC operators, all of it needed to be moved. It was an onion peeling exercise that at times never felt like it was going to end. The documentation associated with these legacy apps/tools was extremely sparse — and that'd be an extremely kind way of putting it. We relied on network sniffers, tribal knowledge, and good ole trial and error to identify strategies as to how to migrate everything that fit under this umbrella.
And even though that was nearly 10 years ago, I’d be willing to bet that if you were to peek under the hood of any enterprise data center today, you’d see roughly the same landscape: a handful of client-facing apps running on the most expensive hardware — with the latest networking, security protections, and protocols — and a bunch of legacy piled neatly in a corner. And that corner would also happen to consume about three-quarters of the data center. This land of forgotten apps, tools, services, and software in most cases often provide the technology backbone that powers most businesses today. Yet at the same time, this big basket of nasty legacy represents the 2nd through the n-th tiers of “software assets” within the enterprise. These systems are old, rarely touched, have dependency trees that would horrify any modern developer, and in most cases, run on extremely dated hardware and OS software.
Well, why is this important? As the curtain is set to open up a second act of the cloud-native revolution within the enterprise, that neatly stacked pile of legacy running in your data center is about to take center stage. In the first act, the 12-factor methodology and the thirst for continuous delivery brought about an architectural revolution that is helping to modernize core systems into cloud-native apps within the enterprise. The business needs have helped to identify these system modernization candidates and placed them at the top of CIOs' priority lists. This modernization is all paid for by the aspirations of expediting the deployment of features from the developer machine to the production server faster than ever before. But as enterprises embarked on their cloud-native journeys and as mission-critical apps were modernized, that foundation of second tier legacy componentry and purchased software — the "everything else" — was still running in the data center. How would the cloud-native revolution account for these apps and services?
As containers and container orchestrators like Kubernetes burst onto the scene in the past few years, containerized workloads have quickly become a real option for enterprises in a cloud-native world. There's no better example of this than this week’s GA release of Pivotal Container Service (PKS). As cloud-native PaaS platforms began to roll out years ago, Pivotal’s Cloud Foundry (PCF) product quickly rose to become the enterprise’s choice for attaining speed and achieving continuous delivery in a newly formed cloud-native world. The platform also helped enterprise IT refocus the technology organization’s time and effort to stay above the value line in order to help their business partners achieve the outcomes they desired. And while extremely successful, Pivotal customers wondered where containerized workloads, especially with all of that legacy still running in the data center, would fit in.
Last December, I had the opportunity to attend Pivotal’s official global partner event, a day before Pivotal’s annual SpringOne Platform event, and the most important announcement was the elevation of containers and Kubernetes as first-class citizens on the PCF platform via PKS. I immediately saw the path for Pivotal customers and all the legacy running in their data centers.
PKS will provide a comfortable environment to run freshly containerized legacy workloads and packaged software in an on-premise model to start, if desired, on top of VMWare vSphere. Access to legacy services and dependencies may prove to be critical early enablers as enterprises embark on moving these older apps and services from servers, where they have resided — literally — for years. Also important will be PKS’s ability to monitor and update Kubernetes via Cloud Foundry’s infamous BOSH interface. PKS also provides VMWare NSX-T technology integration which allows operators to programmatically manage software-defined virtual networks, which will keep your Kubernetes environment secure. In general, platform operators will find a familiar and pain-free way to run containerized workloads at scale on PCF via PKS.
And as the second act of the cloud-native revolution plays out, expect to see more open source innovation to take place around Kubernetes and containers. Companies like Google and Pivotal have a unique mindset in that open source innovation and collaboration will help form a rising tide that will lift boats in the water. As time goes on, we expect to see more enterprises align with this model of thinking.
If you are interested in how to containerize the "everything else" running in your data center and get it up and running on Pivotal Container Service (PKS), please let me know.