Konvoy 1.3 incorporates a number of new features and capabilities that increase the flexibility, governance, and productivity of the product, while continuing to provide you with everything you need to simplify the experience of running Kubernetes on your chosen infrastructure—public cloud, virtual machines, or bare metal.
3 min read
- Air gap support: To broaden applicability beyond on-prem, AWS and other cloud provider support, we have added the ability to implement Konvoy in an air-gap setting. We have customers successfully running Konvoy in air-gapped environments both on-prem and in the cloud, combining the agility of cloud native with the security of isolation from public networks.
- Simplified Azure provisioning: you can now spin up an Azure cluster with a simple “konvoy up” command, increasing your ability to move seamlessly across platforms
- GPU support: Specialized GPU-based processors are playing an ever-increasing role in AI, machine learning and other forms of big data analytics. D2iQ has a deep history of supporting stateful workloads in the cloud native setting. With this new release, you can set GPUs within a specific node and affiliate the workloads best fitted to GPUs based on taints and tolerances. We are working on operators that take full advantage of the capabilities GPUs provide for specialized workloads and will be releasing them in coming quarters—stay tuned!
- Plan-only preview of infrastructure changes: we have added the ability to see what the infrastructure effects of provisioning changes will be before you actually make them, thus reducing the potential for unpleasant surprises or unintended consequences
- Automated custom certificate support: we have added, based on market demand, the ability to automate custom certificate support via the cluster.yaml file, strengthening the governance capabilities of the distribution
- Google Cloud support (private beta): along the same lines as the increased ease of using Azure as your infrastructure of choice, Konvoy 1.3 will enable those participating in the beta program the ability to easily provision clusters on this platform, broadening the range of options you have available for finding the best place to run your applications
- Finally, we are changing our default Kubernetes version to 1.16, to incorporate all the great additions that have been made to core Kubernetes and to better support KUDO
- 500 master and 2500 executors of Jenkins
- 1000 concurrent Spark jobs launched via 50 operator instances.
- Terasort benchmark in ~22 minutes
- 4.6+ m Kafka messages per second with 300+M over life of test
- ~260K writes/second on Cassandra with the expectation we can easily reach 1m writes/second