Achieving Fast DevOps at High Scale with Managed Kubernetes

Kubernetes has emerged as an effective technology for managing containerized workloads across platforms, environments and geographies, greatly streamlining operations workloads. More specifically, Kubernetes has helped organizations achieve fast DevOps. How so? That’s the question Jagmeet Chawla with The Weather Company and Jason McGee with IBM Cloud Platform answered during the keynote address at ONUG’s Fall Conference. 

The Weather Company is the number one digital weather property in the world, claiming 30 million active users through weather.com and wunderground.com during normal weather and two to three times that amount during severe weather. They are producing between 25 and 50 billion forecasts per day.  Chawla spoke about the company’s journey to IBM cloud and how managed Kubernetes is helping them achieve fast DevOps. 

The Mission

Sixty-two miles separate us from space. Collecting data and mapping that critical area provides significant opportunities in many areas. The Weather Company is specifically obsessed with powering billions of personalized weather forecasts each day. This breakthrough has the potential to drive better decision-making that would result in growing retail, saving energy, making air travel safer and turning insurance companies into prevention companies. The more we know about those critical 62 miles, the better decisions we can make. Keep people safer. Make businesses smarter.  

In short, the company’s mission is to map the atmosphere. That mission requires solving a four-dimensional problem.  The first and second dimensions come from latitude and longitude, achieved by simply laying a grid on top of nearly 200 million square miles. The third dimension comes from the 62 miles of elevation,  which is measured from sea level to the Karman line. Elevation is a critical variable because it plays an integral role in weather conditions. The fourth dimension is time. Weather is referred to as a deterministic chaotic system because it is always rapidly changing. 

The Data Process

The Weather Company starts with a lot of data coming in from various sources, including weather stations, Doppler radar, the National Weather Service, as well as other variables, such as turbulence. They take all that data and use weather data modeling to produce a forecast. That resulting data is subbed out via APIs to numerous recipients, including weather.com, various industries and technology partners, such as Google, Apple and Samsung. The entire process requires heavy computing, and is managed in the cloud. 

The Journey to IBM Cloud

Moving to IBM Cloud allowed The Weather Company to expand its global regions to five with two availability zones in each, enabling it to more effectively handle their traffic load and reduce latency. They used the move to IBM cloud as an opportunity to modernize their entire technology stack. How so? Previously, their serving stack was based on Drupal and their altering stack was hosted by AWS by Acquia. They containerized both and now fully run on IKS, IBM Cloud’s Kubernetes services. In addition, they converted their web serving stack to Microsoft Assist, containerizing that configuration and are now running it fully on IKS.  They did not simply do a “lift and shift.” They modernized their entire technical stack, paying down their technical debt. 

The Result

The Weather Company completed its migration to IBM cloud in 2018 over a six-month period, with the following assets.

  • 7 Data Centers

  • 5 Global Regions

  • 460 Nodes

  • 8122 Pods

One month after the migration was complete, Hurricanes Florence (September 2018) and Dorian (October 2019) hit. They were able to effectively scale to handle the increased traffic, which peaked at 2.6 times their average requests. They experienced zero downtime and 100 percent end-user access. There was no impact in terms of performance or availability from the end user’s standpoint. In fact, they saw 30 percent better performance in page load time and increased release velocity dramatically. 

During the three-day storm peak, they had 10 releases, and now they’re averaging 40 percent more releases than before the IBM cloud migration. Moving to IBM Cloud Kubernetes made them more scalable. They had zero downtime over the entire year, including during peak storm time. And, there was no customer impact. 

Why They Chose IBM Cloud?

Chawla summed it up in two words, managed services. “IBM Cloud allows us to focus on the product more than the infrastructure,” he explained.  That’s the primary reason they have significant DevOps savings as well. Here’s how their stats worked out.

  • 100 percent uptime 

  • 80 percent reduction in their DevOps pipeline

  • 40 percent increase in releases

Their DevOps team explained that Kubernetes “enables us to concentrate more on doing the job of supporting the weather.” Chawla summed up their decision in these three descriptors of IBM Cloud Kubernetes.

  • Secure 

  • Stable 

  • Scalable

Weather.com, a high-scale, elastic website proves these three qualities to be true. As 2020 progresses, The Weather Company is starting the process of migrating to Red Hat OpenShift. 

IBM’s Perspective

“We feel really strongly that Kubernetes is the platform for people to use going forward both for new application development and for modernizing existing applications and bringing them into the cloud,” explained McGee.  What elements of IBM Cloud contribute to user success?

  • Infrastructure: IBM invested a tremendous amount in building world-class VPC capabilities, new generation infrastructure around 100-gig networking and faster provisioning. Cross-regional data storage is a huge benefit. US companies can have data stored on the West and East coasts, and have access to it anywhere on the continent with low latency. “Even if you had a full region failure, your data is going to be writable and readable anywhere in the world.” 

  • Kubernetes platform: OpenShift allows companies to have a common platform for containers that work with not just IBM Cloud, but anywhere they need, including Amazon, Google and Azure. 

IBM Cloud rebuilt all its services on top of Kubernetes. That means 90 percent of all the services in IBM cloud are built on the same service that it makes available for clients. That gives IBM a tremendous amount of operational experience. It is committed to making Kubernetes a platform that can support a huge workload. 

While industry experts debate about what kinds of workloads can run on Kubernetes, IBM is proving it’s ideal for stateless websites, relational databases, data warehouses, machine learning, blockchain, IOT or any kind of workload you can think of. IBM has worked hard to containerize its middleware, creating cloud packs that run in an optimized form on platforms like OpenShift, making it easy to deploy IBM Cloud. 

IBM has built its cloud platform on the idea that most companies are operating in a hybrid environment. It plays a fundamental role in providing a consistent Linux platform and a consistent container platform with OpenShift. All capabilities involving data,  AI, applications and integrations hinge on managing that multi-cloud environment. IBM wants to provide that consistent hybrid platform. IBM Cloud stands out in three ways:

  • Open Innovation

  • Security Leadership 

  • Enterprise-Grade

IBM’s goal is to help clients be successful in running their platform anywhere they need it. 

Join Us This Spring!

ONUG is hosting another informative event on building, running and securing the enterprise cloud, May 6 and 7 in Dallas. Don’t miss it! Learn more here or contact us.

Author's Bio

Nick Lippis

Nick Lippis

Co-founder and Co-chairman, ONUG