Since we released 1.0 back in July, we’ve done a lot of work to help people get into production. Not surprisingly, we had to do some patch releases (6 so far!), but we’ve also been hard at work adding new features to the product. The theme for 1.1 is Enterprise Ready.
We’ve been very pleased to see more and more companies using Istio in production, but as some larger companies tried to adopt Istio they hit some limits. One of our prime areas of focus has been performance and scalability. As people moved into production with larger clusters running more services at higher volume, they hit some scaling and performance issues.
The sidecars took too many resources and added too much latency. The control plane (especially Pilot) was overly resource hungry. We’ve done a lot of work to make both the data plane and the control plane more efficient.
You can find the details of our 1.1 performance testing and the results in our updated performance ans scalability concept. We’ve done work around namespace isolation as well. This lets you use Kubernetes namespaces to enforce boundaries of control, and ensures that your teams cannot interfere with each other.
We have also improved the multicluster capabilities and usability. We listened to the community and improved defaults for traffic control and policy. We introduced a new component called Galley.
Galley validates that sweet, sweet YAML, reducing the chance of configuration errors. Galley will also be instrumental in multicluster setups, gathering service discovery information from each Kubernetes cluster. We are also supporting additional multicluster topologies including single control plane and multiple synchronized control planes without requiring a flat network.
Source: istio.io