Conquer Downtime: The Power of Three in High Availability

At CloudBees, we understand that managing downtime and optimizing resource utilization on a large scale can be like turning a tanker ship. It takes time and planning to do it correctly! That's why we continually look for intelligent solutions to manage your infrastructure and keep your productivity sailing smoothly.

In September, we introduced High Availability (HA) for Cloudbees CI, minimizing downtime and keeping services up and running for those who depend on them. This is a huge jump in performance and scalability for Jenkins, and addressed the instability and MTTR pains we have heard from many admins - particularly for larger organizations with large Jenkins installations. But we are not done yet. What have we been up to lately to continue to shave off even more downtime for you?

We are excited to introduce three new powerful features added to our High Availability (HA) functionality: Explicit Load Balancing, Rolling Upgrades, and Hibernation for HA-enabled controllers. 

For the high-availability architecture, the load balancer spreads the workload between the different controller replicas. Explicit load balancing solves the challenge of intelligently distributing builds to the replicas. By tracking internal metrics such as the number of running builds, their estimated remaining durations, and the total connected agent count, the system can elect an owner from among replicas with a relatively low "load." This ensures that builds are scheduled on the replica with the least load, improving performance and efficiency. Read more in the documentation.

Video coming soon to CloudBeesTV

Next, our Rolling Upgrades feature is set to address a critical challenge – Downtime During Upgrades. 

We understand the frustration of downtime during upgrades and introduced rolling restarts to handle infrastructure upgrades like weekly Kubernetes updates with our initial offering of High Availability Mode. But, now is the time to look at our own CI tool upgrades and provide zero downtime solution for this routine. Traditionally, every CloudBees CI upgrade event could cause a halt in productivity. Not anymore. We aim to ensure that your CloudBees CI upgrades happen seamlessly, without downtime, so you don't miss a beat. With our new rolling upgrades feature built into the HA Mode, your CloudBees CI controllers can now be upgraded without disrupting your day-to-day operations. Forget about loss of productivity, with rolling upgrades, your services continue without a hitch, even during an upgrade. Read the documentation to learn more.

Hibernation for controllers is nothing new for CloudBees CI. Hibernation for controllers addresses the cost for idle resources in large organizations. Many managed controllers are heavily used during working hours but sit idle during non-working hours, costing the same even though they are not used. With hibernation enabled, idle controllers can be automatically scaled down to zero replicas, saving costs on infrastructure resources. When activity resumes, the controllers are scaled back up to the desired number of replicas. This optimization of resource utilization reduces cloud spending and infrastructure costs for large-scale organizations. Now, this feature is available to all of your high-availability controllers as well! Regardless of whether you are using HA Mode or not - don’t pay for controllers that are not working for you. Learn more about Hibernation.

Maintaining high availability in an ever-changing digital environment is no small feat. Introducing these three features into our High Availability mode further strengthens our commitment to providing seamless, uninterrupted services to our customers. Let's take on the challenge of downtime together, ensuring your services are always up, always reliable, and always ready to deliver.

Attend the March 6th webinar on DevOps Uncovered: An Interactive Panel on Navigating CloudBees CI and Jenkins, where we will cover these latest features of HA and much more.