DockerCon 2017 was three days of jam-packed action. As each day closed, I took some time to reflect with my colleagues on the hottest elements of our demos and discussions. We ended up with three solid demo videos covering key aspects of our solution for Docker environments.
Auto-discovery and the power of Smartscape
At previous employers, I’ve demonstrated a number of products, but I’ve never focused on the deployment of the product. After all, installation is just a (generally painful) means to an end. But with Dynatrace, deployment is a key part of the story. Not only because it’s so simple, but because it kicks off an automated discovery process that is nothing short of amazing.
In just a few minutes you go from running a command line on your host, to having a fully visualized environment and topology map we call Smartscape. And all using a single agent regardless of your environment or application type.
This is especially key for Docker environments running microservices. When you break up your monolithic app into microservices, the number of services and processes you need to monitor suddenly increases exponentially.
For those attendees who have gone through the time and effort to manually instrument their apps, you can see it dawn on them what this means. Instead of choosing 5% of their critical apps to monitor, why not monitor all of them?
In the video below, Dynatrace SE Peter Hack shows you how it’s done.
One of the best parts of being at DockerCon was showing attendees what Dynatrace does specifically to monitor Docker containers and the microservices running within them. My colleague Scott Kelly wrote a blog sharing some of the reactions, which range from “nobody can do that” to “we didn’t even know that was possible” to “how are you able to do that?”
The traditional approach to monitoring applications running in Docker would be to modify your Docker images to add an agent before running the container. In a large-scale microservice deployment that becomes painful to even consider.
With Dynatrace, you simply install the agent on the underlying host machine (see above). That is the only step required to automatically instrument all the containers, and all the processes, running inside those containers. You can see why people think it’s magic.
This gives you some important metrics specific to your Docker environment. Things like the number of images you have deployed, the number of instances of each image running on which hosts, resource consumption by each container and, of course, the ability to drill into the services running in the containers.
In this video, Dynatrace SE Director Asad Ali shows all the great information you can get from our Docker dashboard.
Finding and resolving problems inside a container
Okay, now we’re deployed (automatically), Docker containers instrumented (automatically) and our environment is fully mapped out (you guessed it, automatically). So how does that help us solve problems?
As you might expect, finding and resolving issues becomes much more challenging in a microservice-based environment. You have many more things to keep an eye on, and they’re all moving. It’s like finding a needle in a haystack, only now it’s a hundred haystacks and they’re in the middle of a tornado.
That’s where artificial intelligence comes in. With the complexity of these environments, it’s no longer possible for a human to make sense of it all. Machine learning and AI are essential to understanding the nature of the problem and identifying the root cause.
When one part of your app stops working or gets overloaded it usually triggers another part to stop working, which then causes two more services to blow up, and so on. We refer to this as “the mushroom cloud effect”. It’s also known to many as an alert storm, because a single problem can trigger hundreds of alerts, and that’s not actually helpful for solving the problem.
Our AI engine has the intelligence to understand that all of these alerts are related to a single problem. Not only that, we also show you the exact location of the root cause. You can even play back how the problem evolved over time.
It’s pretty cool, but better when you see it. Our SE Matt Miller walks through it in the last video.
Thanks to Docker for a great event
With DockerCon two weeks behind us now, I spoke with one of our sales reps who is following up on leads from the show. She said the interest level has been amazing, with several prospects wanting to schedule follow up meetings to dive in further.