Looking at the Manager’s Manager: Instana and Rancher
Portworx sponsored The New Stack’s coverage of KubeCon+CloudNativeCon North America 2019.
When something goes wrong, you don’t just need access to logs. You need the whole picture.
The need for context and complete visibility into the application structure led application performance management company Instana to build an integration with Rancher, which they announced during KubeCon in San Diego.
“They’re almost like a manager of managers for Kubernetes,” explains Pete Abrams, chief operating officer at Instana. “It’s a slightly different take from say, an EKS or GKS. And their approach to Kubernetes is almost identical to our approach to APM.”
What he means is that there are two primary, complementary use cases for both Instana and Rancher — a developer use case, which is about getting quick feedback (in Instana’s case) and setting up a Kubernetes environment quickly (in Rancher’s case). Then there’s the operational use case: Ensuring stability, quality and security, as well as giving operators all the context they need to solve problems quickly when they do occur.
But there were other reasons to support Rancher specifically. Instana was seeing customers using Rancher to deploy Kubernetes environments and using Rancher to manage configurations, but didn’t have visibility to zero in on problems when they happened as a result of configurations Rancher was managing.
“Rancher is a layer over and around Kubernetes,” Abrams says. Instana already had support for Kubernetes and was already able to tell, for examples, what flavor of Kubernetes was being used and to track problems specific to a certain Kubernetes distribution. Before the Rancher support, however, Instana couldn’t tell if Rancher was used to deploy those environments or not.
“It’s all about context,” Abrams says. “There was no connection back to the fact that Rancher was used to create that Kubernetes environment.” Essentially, there was a part of the stack that wasn’t visible, making troubleshooting more difficult and time-consuming.
“Let’s say I have a performance issue, and I figure out that the performance issue is because there’s not enough processing capacity attached to that service,” Abrams says. “Well, wait. Where are the configuration rules that dictated how that environment was created?” If you can immediately see that it’s not in GKE, it’s in Rancher, that shortens time to resolution.
“It’s the contextual awareness,” Abrams says. “That’s what people need when things start acting strange. They need to understand when this thing’s connected to that thing and it depends on that thing. Now that’s more readily at my fingertips.”
KubeCon+CloudNativeCon is a sponsor of The New Stack.
Feature image via Pixabay.